VideoHelp Forum
+ Reply to Thread
Results 1 to 12 of 12
Thread
  1. New to the video world. I have a GoPro Hero5 with a gimbal and I'm testing it all out. I shot a video of me testing out the gimbal in 24fps as sunset (low light). After hours of editing footage in FCPX and reviewing the footage on YouTube, I feel like maybe I should've shot in 30fps? I'm learning FPS and how it affects video. I feel like the footage isn't as smooth as it could be. Btw, ISO was set to 400 max. Is this footage choppy? And should I've shot in 30-60fps with a higher ISO because I"m moving fast? Thanks. https://youtu.be/DA2rH0iY_U8
    Quote Quote  
  2. That video was converted from 30 fps to 24 fps by discarding every 5th frame. That creates several jerks per second in addition to the flicker/jerkiness inherent in 24 or 30 fps video. Shoot 60 fps if you want smooth video.

    Shoot with the exposure set as close to 1/fps as you can. Ie, if you shoot 30 fps try to set the exposure to 1/30. That will get you motion blur which helps reduce flicker.
    Last edited by jagabo; 20th Mar 2017 at 07:35.
    Quote Quote  
  3. Far too goddamn old now EddyH's Avatar
    Join Date
    Jan 2003
    Location
    Soul sucking suburbia! But a different part since I last logged on.
    Search Comp PM
    Heh, cute pup

    I wouldn't really call that "low" light, certainly not for a GoPro type camera, if it's worth the money paid for it. They should be quite good at least into dusk or from mid dawn. I'd certainly expect both my cheapie car dashcam and my phone's camera (it's a Galaxy S5 Mini, not really bought as a digicam replacement, I have a separate one and it scored poorly on that particular front) to work perfectly fine with default settings in the conditions on display here - the sun's still up and shining straight into the lens at times, after all! You should be able to get away with at least 30fps if not 60... also I wouldn't bet against the camera reporting that the video was recorded with ISO 400 being the maximum used/allowed, but actually having used a somewhat lower setting most of the time. A typical cheap digital stills camera could probably manage an exposure somewhere between 1/20 and 1/100s at ISO200 in those conditions depending whether in a fully sunlit position or deep shade, and even 1/20 is often enough for a "steady" image without any in-camera stabilisation if you don't have shaky hands (and are standing still!), let alone using a gimbal. For video, pushing that to 1/30s with a maximum of ... ISO 300? (I'm not entirely sure how to work THAT out), or an occasionally grainy or inter-frame-blurred 1/60s at upto ISO 400 (but mostly less) should look entirely fine and in fact easier on the eye than an absolutely temporally crisp version with no motion blurring at all (which is, e.g. why some of Michael Bay's Transformers action sequences are so utterly impossible to follow, as you lose all the subtle cues that let your eyes track the movements properly).

    That said, I played parts of the video back on both of my desk monitors, one of which has a 75hz refresh and the other 60 despite both being LCDs (I'm in a PAL region, so the slightly more 50fps and much more 25fps compatible scan rate is useful for playing regional footage smoothly, alongside NTSC stuff on the other), and it doesn't seem super-choppy on either of them. There's a little unevenness here and there, but it's no worse than I'd typically expect from youtube videos or a slapdash bit of portable camera conversion on a TV news bulletin. Really, 24fps should be avoided unless you have a good reason for using it (like you want the lower framerate effect and will only be showing the result on a Film-24 compatible hi-def display), and it won't really have gained you a great deal of extra light energy per frame (like 25%, so the rough equivalent of ISO 500 sensitivity whilst retaining ISO 400 grain) but it hasn't gone THAT badly here.


    If the light's really low and the camera supports it (I know *some* action cams do, and really if any would it should be GoPro), it might be worth investigating 20 or 15fps, so long as that actually bumps up the length of time the shutter is open for. The decimation from 60 (at 20), or 60/30fps (at 15) will be noticeably silent-movie-ish, but the inherent motion blur should help cover some of that up and make it more of a stylistic thing. You could probably up the resolution as well by doing that (e.g. to 4k instead of 1080p or whatever), and use a fairly soft resampling resize filter back down to HD or even SD resolution to ramp up the effective ISO and/or brightness by combining/averaging 4 (or 16~20) original pixels down into one.

    A maximum of 400 ISO seems really poor, btw. I would have thought the inherent temporal noise-averaging effect of video would mean it could be possible to get away with 800 or even 1600, to catch grainy but usable low-light footage that could later be processed into something slightly less gnarly, even if still shots would already show some degradation from 100 to 200 to 400... I've rescued a surprising amount from what initially looked like almost completely black videos before, with a combination of extreme level boosting and then pushing the noise reduction about as far as it could be taken. The result wasn't exactly pretty, but it did at least illustrate some of what was going on, and does mean that an initially sort-of-dark recording has a lot of scope for touching up before things get ugly.

    (on that front, I think there are some temporal smoothing filters that could also help reduce the overall noise level of non-moving parts of the footage by essentially applying multi-frame motion blur, and maybe a smaller amount to slower movements, whilst allowing fast motion to remain sharp and free of blur-trails albeit noisy - which shouldn't be as noticeable there - by not smoothing those parts... similar concept to motion detecting deinterlacers, or indeed MPEG compression itself... which would mean you could shoot at 30 or 60fps, even if that means the picture comes out a bit dark and needs to be level-corrected later, amplifying the grain already inherent at ISO400 sensitivity, as the more stationary parts that the eye can analyse at leisure will have smoother dynamics, and the actually noisy parts will be moving around quickly anyway...)

    Happy filming ^_^

    Oh, and just keep experimenting. See what happens at all the different framerates and/or resoltuions in the conditions you're likely to shoot, and indeed even if you mess with other settings like ISO, shutter speed (if you can divorce it much or at all from framerate), aperture (probably unlikely), EV correction (I actually have to crank it down a little for my dashcam...), etc. You may find that you get the perfect result with a combination you might otherwise have never considered.
    -= She sez there's ants in the carpet, dirty little monsters! =-
    Back after a long time away, mainly because I now need to start making up vidcapped DVDRs for work and I haven't a clue where to start any more!
    Quote Quote  
  4. Ah, how did you know that? I just checked and just the first few clips were 24fps so that is what it selected by default. The rest of it is actually 30fps. So... I guess i can't retroactively add frames back? i don't know how it works. Would redoing the project to 30 fps make any difference? And I guess I'd have to re-edit and start all over if I wanted to. Thanks for getting back.

    Originally Posted by jagabo View Post
    That video was converted from 30 fps to 24 fps by discarding every 5th frame. That creates several jerks per second in addition to the flicker/jerkiness inherent in 24 or 30 fps video. Shoot 60 fps if you want smooth video.

    Shoot with the exposure set as close to 1/fps as you can. Ie, if you shoot 30 fps try to set the exposure to 1/30. That will get you motion blur which helps reduce flicker.
    Quote Quote  
  5. Haha, thanks. Appreciate it. Thanks for the reply. So much helpful info in there. I'm only just learning and a lot of the youtube videos ive seen say to set it at 400iso. Im thinking that was in even lower light though. I should've at least selected 800. And you're right, it wasn't really 'low light". If you choose ProTune on the gorpo you can set a lot of the settings manually just like you would with a DSLR in manual mode. I just dont know how that translates to video, i.e. frame rate vs shuttter speed and what it all looks like. so ill do as you said and experiment. Thanks for the tips!

    Originally Posted by EddyH View Post
    Heh, cute pup

    I wouldn't really call that "low" light...
    Quote Quote  
  6. Originally Posted by jaebird82 View Post
    Originally Posted by jagabo View Post
    That video was converted from 30 fps to 24 fps by discarding every 5th frame.
    Ah, how did you know that?
    Stepping through the video frame by frame I could see that every 5th frame was missing -- ie, there was a jump in the motion. Also, at normal playback speed you can see the little jerks that creates.

    Originally Posted by jaebird82 View Post
    So... I guess i can't retroactively add frames back?
    There are techniques to do so but it's rather difficult. Attached is a segment of your upload where I inserted motion interpolated frames for the missing frames. I used AviSynth.
    Image Attached Files
    Quote Quote  
  7. I do see the difference. So that's what it would look like if i initially created the project at 30fps? It looks like i'll be redoing the entire project in my timeline only because I can use the practice anyway. Ill just create a new project and try and match my cuts from my previous one or maybe add different footage to switch things up. Thanks for your help.

    Originally Posted by jagabo View Post
    Originally Posted by jaebird82 View Post
    Originally Posted by jagabo View Post
    That video was converted from 30 fps to 24 fps by discarding every 5th frame.
    Ah, how did you know that?
    Stepping through the video frame by frame I could see that every 5th frame was missing -- ie, there was a jump in the motion. Also, at normal playback speed you can see the little jerks that creates.

    Originally Posted by jaebird82 View Post
    So... I guess i can't retroactively add frames back?
    There are techniques to do so but it's rather difficult. Attached is a segment of your upload where I inserted motion interpolated frames for the missing frames. I used AviSynth.
    Quote Quote  
  8. Originally Posted by jaebird82 View Post
    So that's what it would look like if i initially created the project at 30fps?
    Yes -- except for the distortions that motion interpolation sometimes creates. Step through the video and you'll see them, especially near the edges of the frame. See the bottom of frame 139, for example. Compare it to the frame before and the frame after.

    Image
    [Attachment 40997 - Click to enlarge]
    Quote Quote  
  9. Very interesting. You are really good at explaining things and dumbing things down without sounding too tehcnical btw. Thanks again.

    Originally Posted by jagabo View Post
    Originally Posted by jaebird82 View Post
    So that's what it would look like if i initially created the project at 30fps?
    Yes -- except for the distortions that motion interpolation sometimes creates. Step through the video and you'll see them, especially near the edges of the frame. See the bottom of frame 139, for example. Compare it to the frame before and the frame after.

    Image
    [Attachment 40997 - Click to enlarge]
    Quote Quote  
  10. Far too goddamn old now EddyH's Avatar
    Join Date
    Jan 2003
    Location
    Soul sucking suburbia! But a different part since I last logged on.
    Search Comp PM
    Ah, if you're still learning how all these things go together, then I might suggest putting the camera in automatic mode and seeing how things come out. I expect the vast majority of their customers do exactly that and most of the results are fairly good.

    Video is a little more forgiving than still photography, oddly enough, because you can typically get away with longer exposures and even if it gets a bit blurry from motion and unsteady hands from time to time (aka "shakycam" when used in cinema, as opposed to the more typical Steadicam™ ), generally it will calm down to a more stable and less rapidly moving shot within a few seconds and the viewer can get their bearings again. So you don't need to specify everything super exact unless you literally are shooting a Hollywood blockbuster and have thirty years of experience in the biz.

    (in fact, a little motion blur is a good thing ... seen an action movie recently where the fast combat scenes were rather choppy and hard to follow? that's because some idiot bandwagon-jumping director or camera manager responsible for the chop-socky bits decided to follow the super annoying trend of turning the exposure speed way up so that the sequence looks like a flickbook or one of those old what-the-butler-saw handcranked zoetrope things... a series of overly sharp still photos with no overlap, or any inherent motion information of their own, so the ultimate result is that of a cartoon where the animator didn't put in any speed lines or bother to motion-distort the moving items, and your brain ends up having to work far too hard trying to integrate each 1/24th sec flash of new information instead of it flowing together more naturally... it's hard to make sense of things and is generally tiring. Do you *see* the world that way? No, of course not...)

    Somewhere I had a nice instructive picture that I'll upload if I can find where it went, showing the effect of changing each setting in a pictogrammatic form...

    But in the absence of that...

    Exposure time: in a normal camera, how long the shutter is open for, letting in light to fall on the film. In a shutterless digital camera, how long the CCD is strobed active for, collecting charge in its matrix of receptor cells from the action of photons falling on them. Obviously in a movie or video camera this can't be longer than the framerate, though the electronic version does at least have the advantage of a very short gap between the end of one exposure and the start of the next, which could be quite substantial with film (as it just needs to read and discharge the CCD instead of closing the shutter, moving the film, opening it again... and evidently as some very high frame rate recording options exist, the read/discharge cycle can happen very quickly indeed).

    You need this to be sufficiently long that enough photons can fall on the film or the CCD for a clear image to form (typically the larger and/or lower resolution the film or sensor, the less time this takes) with a good contrast between light and dark, at least some in-between tones, and sharp edges, without being so long that either excessive motion blur / camera shake becomes a problem for image clarity, or the image starts to over-expose (the film or sensor reaches the limit of how much light energy it can respond to in a single spot, or possibly across the entire frame). Usually unless you have a good reason to fiddle with it, I would say to let the camera decide. Any decent film camera (except the simplest old mechanical point-and-shoots which may have had a simple bright/dull/flash-assist switch, or nothing at all...) should have a light meter built in, and every digital camera is its own light meter thanks to the system being able to constantly monitor the peak and average light energy falling on the sensor several times per second before you press the shutter (and video cameras can determine it on a frame-by-frame basis as a side effect of the encoding process). Therefore if what you want is a relatively steady, well balanced amount of light across the scene, the "person" best equipped to decide the correct exposure is... the camera itself.

    There ARE reasons to deliberately adjust the exposure, but they're limited to particular situations - starlight images, synching to a particular TV, movie screen or flickery lightbulb frequency, forcing "slow sync" with flashbulbs at a paparazzi gathering, forcing blur for special effects, forcibly crisping up movement or eliminating camera shake, freezing motion that would otherwise be too fast to see (particularly in conjunction with a strobe light), etc... Unless you have a particular one you need to use, or a max/min that shouldn't be violated, it's better to let it run free.

    Measured in fractions of a second because exposures greater than about 1/15th of a second are unusual (let alone over a second), and it's quite easy to think of the increasing number signifying a faster (virtual) shutter action. In fact you might even see some 1-and-a-bit-second exposures noted as 1/0.8, 1/0.6 or similar (1/0.5 and beyond just become regular "2s", "5s" etc)

    Lens/Shutter Aperture: In essence, how large the hole is that lets light into the camera. Either a separate item that's part of the lens assembly, or a feature of the shutter itself if it's of a type that opens into a circular shape. An important counterpart to Exposure and actually something you might have more reason to actively mess with.

    You'd think that opening it as large as possible to allow the largest amount of light in would be the best idea at all times, but no - both for reasons of potential overexposure with deliberately slower shutters, and because, as demonstrated with the difference between holding a sheet of paper at the back of an open-sided box vs one that's been turned into a Pinhole Camera, the size of the hole has an important effect on the camera's ability to focus... or more precisely, to maintain sharp focus over a reasonable range of differences simultaneously (known as depth of field). I believe it's something to do with diffraction and may well be quantum, but that's not important right now.

    Ignoring the physics behind it, a wider hole = more photons coming in but with a fuzzier relationship between where they come from outside and where they end up inside (and narrower = less light, but more sharply defined over a wider range). Lenses can help with this, but only up to a certain point - really what they do is shift where the centre of the focal range is (ie the focal point) closer or further away from the aperture itself, and the depth of field is (if I remember this right) a certain percentage of that distance, so it's easier to get a good shot of a distant object than a close-up one, and you have to close up somewhat as you get closer (?). The distance between the aperture and the film or sensor is also involved but if the device is well made then the two should be fairly well matched. Possibly it may mean you can get away with a wider aperture if you have a larger sensor (and the reverse) but not sure.

    Do bear in mind my own grasp of this bit isn't the clearest, because one thing I do know is that apertures tend to close up automatically on superzoom cameras when you zoom in, so maybe either I got that backwards, or the zooming function makes it as if the distant objects were now all a lot closer?

    In any case, the relationship between exposure and aperture is essentially one of balancing how sharp or soft a depth of field effect you want (making the DoF quite narrow, in order to focus particularly on a single person or object whilst blurring out everything that's nearer or further is a common in-camera special effect, but at other times you may need the image to be crisp over a great range of distances) with how much light you need to get into the camera, and how fast or slow you want the (virtual) shutter to open and close, with an extra flavour of the actual distance you need to focus at.

    Generally, however, modern lens systems - and the relatively narrow maximum apertures of typical consumer cameras - mean that for all but fairly challenging conditions (macro/extreme close-ups, very bright lighting with longer exposures, extreme optical zoom) the convention seems to be to lock the aperture open at the maximum available and vary the exposure (and ISO) to suit, only "stopping down" when those particular challenges need to be met. If you want to force a tighter aperture for whatever reason, or indeed prevent it going smaller (mechanical/preprogrammed zoom adjustments excepted), then you can take manual control of that and let the camera automatically vary the other settings to compensate.

    The setting is generally measures in "f-stops", where F stands for F***ed If I Know, and, as per exposure, the number given is actually a divider of some kind, such as "f/2.6". Wikipedia probably has a very good article on it that I could look up, but eh. I presume f/1 is an aperture equal to the size of the film/sensor and everything past that is a hole with diameter 1/x that size, something like that - certainly, a larger number in this case means a SMALLER hole (or, if you need a more logical relationship, a wider DoF). The relationship between f-number and amount of light coming in is a bit odd, something like every 1.5 represents a doubling (or halving) of the illumination? Again, I forget, but unless you're going to use a fully manual camera or lock off every single setting without checking what the result is like, it's not massively important at this point... just low number = brighter, fuzzier, high number = darker, sharper. And generally anything below f/2 represents a system that's pretty good at capturing light, especially if that's a "whole system" rating that includes the filtering effect of the lenses (which again can rate as much as that all by themselves at minimum zoom, and more when zooming in, on top of the aperture's own light rejection figure... even optical-grade glass that's only a millimetre thick can absorb a surprising amount of light, if the job is to transmit as much light through it as possible; the system in a pair of binoculars, for example, is considered pretty good if it only darkens the image by 15%), and you may see interchangeable lenses for SLRs described as having a certain f-rating... the lower the number, the less light they (inadvertently) filter out. At the other end, apertures can get quite pinhole-like, with f/11 being common and f/16 or f/22 not being unknown in extreme cases (deeeeeeep-zoom lenses and so-on; you need huge light-concentrating inlets and a nice big sensor to make the most of them).

    ISO number: This is a bit of an odd one... essentially it's an old measurement of film sensitivity (and, quite often, graininess, as a more sensitive film tended to have larger pigment grains unless it was a very fancy formulation). Again, I'm not sure of the exact definition, but I have a feeling it was something like ISO-200 is a medium speed film with general purpose grain, that should be alright for shooting in moderate sunlight at 1/200th of a second. In this case the number is a reasonably direct measure of sensitivity - the higher the number, the more sensitive the film, and the faster you can shoot whilst still getting a similar-looking (though somewhat "rougher", grainier, lower resolution) picture out the other end when processed under the same conditions (actual photo film can be "pushed" or "pulled" somewhat during processing to compensate for under or overexposure, during the secondary part where the developed negative is itself exposed onto the final print paper, rather like adjusting the RAW version of a digital photo in a retouching program - but in both cases there is a limit at each end where the details disappear into the shadows or highlights and can't be recovered, and that's essentially what's governed by the rating). Double the ISO, double the sensitivity, half the exposure time (or aperture area?) for a similarly "lit" image, though you probably lose half of the effective resolution (as in, 1/sqrt(2) in each direction, or maybe 1/cube root(2) in each spatial dimension, and also in the dynamic range one too).

    So, for shooting mostly well-lit scenes with a relatively slow shutter and wide aperture, but capturing a lot of fine detail in the pictures you take, you'd want ISO 100 film, maybe even lower (professional portrait photographers probably use ISO 80 or less, and relatively large-format film instead of plain 35mm?). For taking fast-action sports shots on an overcast autumn afternoon in an urban area, with not very good floodlights on the field, or a lot of indoor/night time shots with a weedy point-and-shoot flash, ISO 400 or possibly higher (I think most single-use cameras are loaded with 35mm ISO-400, and a lot of smaller-frame formats like APS and 110 cartridge didn't tend to offer anything lower than 200 because their limited light capture area meant sensitivity needed prioritising over resolution, even though that was made even worse by the dimensions). Paparazzi trying to get some kind of printable image of a celebrity, snatched in a sudden moment as they hurried past, presumably used ISO 800 or more with their chunky lenses and huge, extra-bright flashguns, so they could still use fast exposure even at night.

    These types of ratings have carried through into digital photography. I don't know if there's actually any kind of direct way of relating the original International Standards Organisation grades (which presumably relied on calibration with a series of lights of different intensities illuminating a particular test image, and a camera with a fixed aperture and exposure...) into digital camera performance, but the same general concepts persist at least - ISO 200 is the "medium", decently-lit daytime setting, which might show a little grain but usually you wouldn't notice. ISO 100 is lower sensitivity but smoother, for when good lighting is guaranteed, and sometimes you might even get 80 or 64 with some models, which more or less eliminates grain entirely. ISO 400 is a more general purpose cloudy/bright-indoor setting, with some slightly noticeable grain but nothing too bad, and ISO 800 is often the highest a lot of cheap cameras will rate, allowing some modicum of indoor night-time (or streetlight) shooting at slightly extended exposure times without needing a flash, but with pretty obvious grain. Again, better cameras offer 1600, 3200 or more, through a combination of better quality sensors (reducing grain at all ISO levels without sacrificing sensitivity - much like the expensive film did) and simply pushing the post-processing even harder, sometimes colouring in the numbers on the display as a warning to the user that the result might look rather noisy, as the CCD signal level isn't riding very much above background voltage any more.

    My phone camera can do 100 to 800, which is OK I guess. But my old Canon Powershot rangefinder digital stills camera (a category whose demise I mourn) offers 80 to 1600, a 20x range instead of 8x, and although I don't often find much use for the lower end of the range (it does allow a slightly longer exposure at full aperture without blowing out the details, but I'm hard pressed to see a difference in detail), that extra bit of sensitivity at the top end, combined with the better lens and sensor too (and the optical stabiliser system) mean it's a much better machine for taking tricky low-light photos. Unfortunately, neither of them give an option for controlling sensitivity during videos.

    Thus we see that actually the exposure control for a camera is threefold, and can often be a compromise, between the varying demands of shutter speed, aperture width, and sensitivity setting (which digital systems bless us with the ability to alter on-the-fly, instead of being beholden to a single choice for an entire roll of film). Depending on what you're shooting, where and when, and how you want to present it, you may want to tweak one or more of those and then have to compensate with the other(s). Happily, the typical control modes on a modern camera allow for a reasonable amount of control over them... the names sometimes vary but the ones of interest here typically comprise:

    AUTO mode (marked as such, or maybe A) - the camera assumes you want a moderate brightness level across the whole frame, under the normal range of lighting conditions encountered everyday (with very dark or very bright seeing it gradually bias towards under- or over-exposure), and with a moderate shutter speed (typically in the 1/60 to 1/500 range) and a medium ISO, plus as wide an aperture as possible unless zoom/focal distance demands otherwise, and will vary all three to try and achieve this without any of them going wildly off into the woods. Different models seem to prioritise particular things (I've one in particular that is convinced that it's perfectly alright to randomly set really long exposures, down to 1/2 second, before it starts to crank up the ISO...) but in general there should be a tendency towards wider aperture, longer exposure and higher ISO in dark places and the opposite in bright ones, with it automatically adapting between shots or through the run of a video. Auto mode is usually deliberately quite locked down to avoid flooding the newbie with too much confusing info, or to give you a "safe" mode to hand it over to a kid with, so the amount of stuff that you can change is usually very limited, sometimes nothing more than changing the flash between "auto" or "off" (but not "forced on"), setting a self-timer and adjusting resolution.

    PROGRAM mode (or just "P", sometimes confusingly "S") - a slightly more flexible setting and generally what I use most. Aperture and exposure are still automatically controlled within similar limits, but you have more control over various items. Particularly, ISO can be changed (either "auto" - which usually limits the maximum to 400 or so - "auto hi" that extends it further and may limit the minimum, or locked to a particular value) so you can balance the overall feel of the pic and what range of speed/width is likely to be used (the name I expect comes from the old auto-program film cameras that detected the speed rating encoded on the body of a 35mm cartridge with a pattern of conductive and insulating squares and adjusted their operating regieme to suit), but also you have greater control of flash (forced-on, slow sync, and flash strength often available), exposure adjustment to compensate for when the automatic setting is either temporarily or consistently lighter or darker than you'd like (usually either one-half or one-third "EV" stops, with a full EV stop being a doubling or halving of brightness and the range usually +/- 1 or 2 stops from full-auto - it's a somewhat crude thing, but still very useful in extreme conditions, especially where some particular unimportant part of the image is "poisoning" the light meter reading to be too low, or more commonly too high, causing the image to wash out or be lost in shadow... eg a lightbulb in the background, low sun at dawn/dusk... it's particularly good for helping to calibrate dashcams, oddly), colour profiles (usually with an option for manual saturation and contrast too), JPG compression strength, manual/auto/macro focus, focus and exposure metering method, that sort of thing. Essentially what you could do with a midrange / prosumer 35mm pocket or superzoom camera around the turn of the century, with some extra digital bells on.

    APERTURE PRIORITY (or just "A", "As", "Ap" or "F") - like Program, but you can alter the Aperture setting, locking it to a certain value in order to achieve a particular effect, with the other settings (ie shutter speed, and - if not itself locked down - ISO) having to compensate for conditions entirely by themselves. Can't say I use it much myself, but then I don't have a really nice camera where you could use it to get some really good depth of field effects, or really crisp macro shots without having to employ extra long exposure and huge ISO. Note that it can't violate any mechanical requirements and if used with a zoom lens might be mechanically forced upwards (or downwards, depending where you start), and then not adjust back down afterwards without manual intervention.
    One particular feature is that, similar to extra high ISO, the camera usually won't go anywhere near its tightest aperture settings unless it really really has to, so if you twiddle the control in that direction you might see some pretty high numbers that you would never have otherwise seen, eg f/11 and beyond on my Canon (usually it doesn't go much beyond 6)...

    SHUTTER PRIORITY (usually "T", "Tp", "Ts"... for "time", I guess?) - as above, but for shutter speed instead of aperture. Lock the exposure time, and the aperture (and optionally ISO) vary automatically to keep the output usable. The adjustment tends to get even further than aperture, when compared to Auto or Program, in that you can only unlock the very fastest or very slowest shutter settings in either Tp or fully Manual mode, otherwise it tends to stay within a more conservative range that can lead to pictures that don't come out at all even if on a tripod (in low light), or more rarely get completely washed out as it won't snap fast enough. Normally the Canon only allows 2s thru maybe 1/2000th (and will step up the ISO as things get darker, through the 1/100th to 1/15th range to try and keep the speed up), and the cheaper backup Nikon won't go slower than 0.5s, but in Ts and Manual it can do as long as 60s or as short as 1/10,000th (or less, even?), and even the Nikon does 15s to 1/8000... Which allows you to go right from taking starry sky pictures outside to getting up close and capturing the smallish writing on the outside of a lit 100w bulb indoors.

    FULL MANUAL CONTROL (typically "M") - As expected, you take full control of aperture AND exposure (with automatic or fixed ISO), in order to really fine tune a shot to your liking. Sometimes EV is still available too, in case you want to bias the ISO setting somewhat without locking it down, or to adjust the range of the subtle "too dark/too bright" warnings that the light meter system provides in this mode and Ap/Tp (it won't stop you taking a picture, but it will e.g. turn the speed/aperture display lettering red or yellow if it thinks there's not enough light coming in to produce a decent average-brightness image... of course, you might WANT one that's extra bright or extra dark).

    Mind that whilst the other modes could be quite useful for various reasons, I'm not really sure if this is particularly useful for a video, to be honest, as the potential for large changes in brightness could lead to your recording going from very bright to very dark with few identifiable details in either case, unless it's got a genuine and effective HDR mode that can be turned on independently of everything else; often I emulate it a little in the standalone cameras by turning the contrast in their colour settings to minimum, and the saturation down halfway, which gives a much more natural look to everything (essentially that's more or less what the camera sees, unfiltered, and the "normal" setting has both of them slightly overcranked by different degrees to make things look more vibrant at the expense of losing some very bright/dark detail and giving a slightly hyperreal cast to everything), but I still wouldn't expect them to be able to cover the usual range of brightness you could encounter just on a 360 degree spin unless I got the settings just-so before starting the run, and the brightest/darkest images would still be hard to properly make out. Really, it's much better for single photos that you want to achieve a particular effect with, or short stylised video clips either in a studio or a single fixed outdoor angle. For everything else, much better to let the camera vary things a bit more if you're not able to rapidly crank it up and down yourself whilst keeping an eye on the live display.
    (that sort of thing is why the upper end prosumer cameras have multiple dials on them still, even with touchscreens etc, so you can rapidly dial all six of the main controls - exposure, aperture, EV, ISO, zoom and focus - up and down using muscle memory and haptic feedback, without having to hunt through various menus and break concentration to stab away at the screen, making the aim wobble)

    On top of which, for video, there's resolution and frame rate. Hopefully you're more familiar with that....

    Resolution, really you want to set 720p as a minimum these days, 1080p if the device and memory card have the speed, capacity, and battery life to deal with it. Going higher is your own choice, and not universally useful at this point. Though, as with still photos, the higher the original resolution, the more you can crop down and digital-zoom without things becoming ugly and pixelated. If you really, really have to set SD, see if a 50 or 60fps progressive mode is available. Avoid interlaced recording like the plague, it's a 20th century relic, and the fact it got OKayed for use with 1080-line HD shows that there are some sick minds in the world. Use 25/30fps instead of it unless you have some properly fast action to capture and need the additional smoothness - the only reason to record 50i/60i any more is if no other high speed mode is available...

    960p, or 1440x1080 non-anamorphic tend to be 4:3 modes (or what gets called "tall" these days!). If you need the extra frame space, then by all means give it a go, but be aware you'll have to either trim it down or pillarbox it for fullscreen viewing on any modern display. It can be useful to allow a roving window within the larger image area (eg 1280x720 within 1440x1080), especially with a fisheye lens, but that's a more advanced processing technique. (1440x1080p anamorphic is still 16:9, just with fatter pixels; "1440p" is something else entirely, basically twice 720p - aka "2.5k" - and not really a standard).

    Unless it's unavoidable, or you really prefer the low-framerate effect for some reason, I'd say it's currently best to record in 60fps progressive*. You can convert down from it with fairly good results, upconversion is a reasonably mature technology (and, honestly, if you can detect the artefacts at 120+ fps, as well as not feeling slightly nauseous, you're some kind of mutant), and if we ignore the remaining stock of PAL and PC CRTs it's the best match for the native scanrate of the vast majority of modern screens, which all default to 60hz.
    (of course, by this we mean 59.94fps, but that 0.06 discrepancy means a single frame jitter every 16.7 seconds even if "60" means 60.00, so it's not a major problem; most videos that have any kind of editing rather than being straight end-trim-and-upload jobs don't have cuts that last longer than 10-15 seconds, so you can engineer it to avoid skips)

    *(or, if your target display is something receiving a PAL broadcast, 50fps... if there's something that's glaringly obvious and annoyingly ugly, it's content that was originally filmed at 30 or 60fps either because it originates from an NTSC area or the creator was ignorant and/or careless, being shown as part of a 50fps PAL TV stream, after poor quality or basically nonexistent framerate conversion. Herky jerky city. I've even seen this crime committed by brands trying to promote the motion smoothness of their new 200hz TV, or the fluidity of a videogame engine... some people just need slapping)

    30fps, or as previously mentioned 20, 15 or even 12 (and the PAL equivalents down to 12.5) can be acceptable substitute rates for reasons of storage space, extra low light levels, or special "8mm" or other timelapse stylised effects, but you wouldn't want to watch a half-hour film at anything below 24 (or half hour TV show at below 25 or 30). There's a reason these standards became established, and it's because anything slower and more jerksome is just plain uncomfortable. Animation can often get away with 12-15ish fps for most of the action, but only for the characters, with a considerable amount of speedlining, fake motion blur and other "hinting" distortions (and the panning across scenes and backgrounds are still at full framerate) to trick the eye into seeing more actual movement than is present.

    At the other end you can go up to 100/120fps or more, but unless you're going to very definitely show it on a lot of HFR displays and you can demonstrate any kind of perceivable difference in a scientifically valid double-blind study, I'd do that only for the possibility of smooth slowmo (replay at 25-30fps = 1/4 speed...) or selectively adding vector-interpolation-based motion blur to a sequence initially shot with a high enough shutter speed to render everything on screen in pin-sharpness. Ditto 200+fps but even more so, as we're decidedly into the realm of just-in-case slowmo capture, and decimating from that speed back down to 60fps without any kind of adaptive blurring may actually toss away a lot of useful motion information that would otherwise have been blurred into the one frame. Additionally, these higher rates obviously burn through a lot more storage space and battery power (and processing power later on), and can limit you to lower resolutions, so it's yet another balance to be made. Really, 1080p at 60fps gives you plenty of flexibility, and anyone who complains about the jerkiness or blockiness at this point needs to be mercilessly robbed^H^H^H^H^H^H befriended in the strongest possible way as their house must be full of all kinds of amazing stuff. If you think you might need more frames or more pixels for cropping or slowmo effects, go up to e.g. 2.5k/1440p (2560x1440) or similar, and/or 120fps, but try not to be forced down to less than 720p or 30fps as a compromise. And check that your card is fast enough to withstand it before recording anything important!

    Anyway, hopefully that should be some kind of useful rundown. Have fun and happy experimenting, and please go make some more actually interesting and nice-to-look at videos, because there's too much crap on youtube at the moment


    NB one thing I didn't mention above as it's only really relevant to stills cameras, and then typically SLRs or really good compacts (BUT these do get used as video cameras quite a lot, too), is the realm of physical filters; either ones that attach externally onto the lens or the camera body, or sometimes built into the camera itself. Naturally these almost always reduce the light level in different ways, but it can be worth it for a particular effect. External ones come in all kinds of flavours - colour tinting, either overt shades or subtle fill-ins to make overcast skies look more blue or add a fake "sunset" cast to things, micro-etched jobs that add starburst or other effects to the "bokeh" (spots of light that are defocussed into circles due to DoF effects), polarisers that help reduce glare and enhance natural contrast especially where low sunlight or reflections from water are involved, and "NDFs" - neutral density filters that universally reduce the amount of light coming in without tinting or otherwise affecting it, so you can use a longer exposure and/or wider aperture to add deliberate motion-blur and soft-focus effects without overexposure, even though you can't drop the ISO any further. Usually they're rated in terms of how many stops they reduce the scene brightness by, eg a "3 stop NDF", so with otherwise identical settings you can open the aperture up that much or slow the exposure accordingly without changing the brightness of the scene. It can be a particularly popular option for natural scenes or crowds, roads, etc where you want a drawn-out timelapse blur effect (eg that particular one of a stream or waterfall looking like a carpet of gossamer silk is achieved this way, and I've actually managed to produce similar with a compact camera in the height of summer by clamping the lenses of two different pairs of cheap sunglasses to the front with duct tape) during the daytime.

    Also that of bracketing, which is both a method of taking multiple shots in an uncertain lighting condition and picking the best one later, and of combining those shots to produce an HDR image; the camera either automatically, or under the control of a VERY steady-handed user, takes two, three, or more pictures in as rapid succession as possible - one (or more) with a negative EV, optionally one neutral (ie EV 0) and one (or more) positive, to create a series of otherwise near-identical images that have known differences in exposure/brightness between them, and a maximum and minimum value therefore rather extended from the expected norm of EV 0, which allows the camera or a suitable program to align the multiple pictures, adjust their contrast and brightness accordingly so the midtones line up and the very darkest colour = black, very brightest = white, then paint in any parts which clipped to flat black in the darker pictures with detail from the brighter ones, and vice-versa for what clipped to white in the brightest ones, possibly with some local tweaking of levels in different parts of the image to improve overall contrast (else it may just look washed out - though going too far with that effect leads to the weird "oversharpened" or "haloed" look that HDR images can have), to create a final image with a lot more brightness range than would otherwise be possible with the typical dynamic sampling width of a typical CCD. Given that there's a particular difference in energy between the brightest and darkest at a particular ISO (as it's basically based on amplifying the semi-analogue signal by a certain amount...), the more pictures there are, the wider the range can be, and the greater the overlap they can have which improves the quality as well. The most I've usually seen is 5 steps, 1EV apart, but you can have 3 or even just 2 that are each 2EV apart (-2, 0, +2, or just -1 and +1) without it looking too awful. -2 and +2 would probably not work
    (that last one is definitely not practical for video unless you manage to hack a 3D camera to have different exposures from each of two very closely spaced lenses, or have something running at 240fps without much motion of its own, taking a series of 4 shots every 1/60th sec at +1.5, +0.5, -0.5 and -1.5EV...)
    -= She sez there's ants in the carpet, dirty little monsters! =-
    Back after a long time away, mainly because I now need to start making up vidcapped DVDRs for work and I haven't a clue where to start any more!
    Quote Quote  
  11. Far too goddamn old now EddyH's Avatar
    Join Date
    Jan 2003
    Location
    Soul sucking suburbia! But a different part since I last logged on.
    Search Comp PM
    Ah, if you're still learning how all these things go together, then I might suggest putting the camera in automatic mode and seeing how things come out. I expect the vast majority of their customers do exactly that and most of the results are fairly good.

    Video is a little more forgiving than still photography, oddly enough, because you can typically get away with longer exposures and even if it gets a bit blurry from motion and unsteady hands from time to time (aka "shakycam" when used in cinema, as opposed to the more typical Steadicam™ ), generally it will calm down to a more stable and less rapidly moving shot within a few seconds and the viewer can get their bearings again. So you don't need to specify everything super exact unless you literally are shooting a Hollywood blockbuster and have thirty years of experience in the biz.

    (in fact, a little motion blur is a good thing ... seen an action movie recently where the fast combat scenes were rather choppy and hard to follow? that's because some idiot bandwagon-jumping director or camera manager responsible for the chop-socky bits decided to follow the super annoying trend of turning the exposure speed way up so that the sequence looks like a flickbook or one of those old what-the-butler-saw handcranked zoetrope things... a series of overly sharp still photos with no overlap, or any inherent motion information of their own, so the ultimate result is that of a cartoon where the animator didn't put in any speed lines or bother to motion-distort the moving items, and your brain ends up having to work far too hard trying to integrate each 1/24th sec flash of new information instead of it flowing together more naturally... it's hard to make sense of things and is generally tiring. Do you *see* the world that way? No, of course not...)

    Somewhere I had a nice instructive picture that I'll upload if I can find where it went, showing the effect of changing each setting in a pictogrammatic form...

    But in the absence of that...

    Exposure time: in a normal camera, how long the shutter is open for, letting in light to fall on the film. In a shutterless digital camera, how long the CCD is strobed active for, collecting charge in its matrix of receptor cells from the action of photons falling on them. Obviously in a movie or video camera this can't be longer than the framerate, though the electronic version does at least have the advantage of a very short gap between the end of one exposure and the start of the next, which could be quite substantial with film (as it just needs to read and discharge the CCD instead of closing the shutter, moving the film, opening it again... and evidently as some very high frame rate recording options exist, the read/discharge cycle can happen very quickly indeed).

    You need this to be sufficiently long that enough photons can fall on the film or the CCD for a clear image to form (typically the larger and/or lower resolution the film or sensor, the less time this takes) with a good contrast between light and dark, at least some in-between tones, and sharp edges, without being so long that either excessive motion blur / camera shake becomes a problem for image clarity, or the image starts to over-expose (the film or sensor reaches the limit of how much light energy it can respond to in a single spot, or possibly across the entire frame). Usually unless you have a good reason to fiddle with it, I would say to let the camera decide. Any decent film camera (except the simplest old mechanical point-and-shoots which may have had a simple bright/dull/flash-assist switch, or nothing at all...) should have a light meter built in, and every digital camera is its own light meter thanks to the system being able to constantly monitor the peak and average light energy falling on the sensor several times per second before you press the shutter (and video cameras can determine it on a frame-by-frame basis as a side effect of the encoding process). Therefore if what you want is a relatively steady, well balanced amount of light across the scene, the "person" best equipped to decide the correct exposure is... the camera itself.

    There ARE reasons to deliberately adjust the exposure, but they're limited to particular situations - starlight images, synching to a particular TV, movie screen or flickery lightbulb frequency, forcing "slow sync" with flashbulbs at a paparazzi gathering, forcing blur for special effects, forcibly crisping up movement or eliminating camera shake, freezing motion that would otherwise be too fast to see (particularly in conjunction with a strobe light), etc... Unless you have a particular one you need to use, or a max/min that shouldn't be violated, it's better to let it run free.

    Measured in fractions of a second because exposures greater than about 1/15th of a second are unusual (let alone over a second), and it's quite easy to think of the increasing number signifying a faster (virtual) shutter action. In fact you might even see some 1-and-a-bit-second exposures noted as 1/0.8, 1/0.6 or similar (1/0.5 and beyond just become regular "2s", "5s" etc)

    Lens/Shutter Aperture: In essence, how large the hole is that lets light into the camera. Either a separate item that's part of the lens assembly, or a feature of the shutter itself if it's of a type that opens into a circular shape. An important counterpart to Exposure and actually something you might have more reason to actively mess with.

    You'd think that opening it as large as possible to allow the largest amount of light in would be the best idea at all times, but no - both for reasons of potential overexposure with deliberately slower shutters, and because, as demonstrated with the difference between holding a sheet of paper at the back of an open-sided box vs one that's been turned into a Pinhole Camera, the size of the hole has an important effect on the camera's ability to focus... or more precisely, to maintain sharp focus over a reasonable range of differences simultaneously (known as depth of field). I believe it's something to do with diffraction and may well be quantum, but that's not important right now.

    Ignoring the physics behind it, a wider hole = more photons coming in but with a fuzzier relationship between where they come from outside and where they end up inside (and narrower = less light, but more sharply defined over a wider range). Lenses can help with this, but only up to a certain point - really what they do is shift where the centre of the focal range is (ie the focal point) closer or further away from the aperture itself, and the depth of field is (if I remember this right) a certain percentage of that distance, so it's easier to get a good shot of a distant object than a close-up one, and you have to close up somewhat as you get closer (?). The distance between the aperture and the film or sensor is also involved but if the device is well made then the two should be fairly well matched. Possibly it may mean you can get away with a wider aperture if you have a larger sensor (and the reverse) but not sure.

    Do bear in mind my own grasp of this bit isn't the clearest, because one thing I do know is that apertures tend to close up automatically on superzoom cameras when you zoom in, so maybe either I got that backwards, or the zooming function makes it as if the distant objects were now all a lot closer?

    In any case, the relationship between exposure and aperture is essentially one of balancing how sharp or soft a depth of field effect you want (making the DoF quite narrow, in order to focus particularly on a single person or object whilst blurring out everything that's nearer or further is a common in-camera special effect, but at other times you may need the image to be crisp over a great range of distances) with how much light you need to get into the camera, and how fast or slow you want the (virtual) shutter to open and close, with an extra flavour of the actual distance you need to focus at.

    Generally, however, modern lens systems - and the relatively narrow maximum apertures of typical consumer cameras - mean that for all but fairly challenging conditions (macro/extreme close-ups, very bright lighting with longer exposures, extreme optical zoom) the convention seems to be to lock the aperture open at the maximum available and vary the exposure (and ISO) to suit, only "stopping down" when those particular challenges need to be met. If you want to force a tighter aperture for whatever reason, or indeed prevent it going smaller (mechanical/preprogrammed zoom adjustments excepted), then you can take manual control of that and let the camera automatically vary the other settings to compensate.

    The setting is generally measures in "f-stops", where F stands for F***ed If I Know, and, as per exposure, the number given is actually a divider of some kind, such as "f/2.6". Wikipedia probably has a very good article on it that I could look up, but eh. I presume f/1 is an aperture equal to the size of the film/sensor and everything past that is a hole with diameter 1/x that size, something like that - certainly, a larger number in this case means a SMALLER hole (or, if you need a more logical relationship, a wider DoF). The relationship between f-number and amount of light coming in is a bit odd, something like every 1.5 represents a doubling (or halving) of the illumination? Again, I forget, but unless you're going to use a fully manual camera or lock off every single setting without checking what the result is like, it's not massively important at this point... just low number = brighter, fuzzier, high number = darker, sharper. And generally anything below f/2 represents a system that's pretty good at capturing light, especially if that's a "whole system" rating that includes the filtering effect of the lenses (which again can rate as much as that all by themselves at minimum zoom, and more when zooming in, on top of the aperture's own light rejection figure... even optical-grade glass that's only a millimetre thick can absorb a surprising amount of light, if the job is to transmit as much light through it as possible; the system in a pair of binoculars, for example, is considered pretty good if it only darkens the image by 15%), and you may see interchangeable lenses for SLRs described as having a certain f-rating... the lower the number, the less light they (inadvertently) filter out. At the other end, apertures can get quite pinhole-like, with f/11 being common and f/16 or f/22 not being unknown in extreme cases (deeeeeeep-zoom lenses and so-on; you need huge light-concentrating inlets and a nice big sensor to make the most of them).

    ISO number: This is a bit of an odd one... essentially it's an old measurement of film sensitivity (and, quite often, graininess, as a more sensitive film tended to have larger pigment grains unless it was a very fancy formulation). Again, I'm not sure of the exact definition, but I have a feeling it was something like ISO-200 is a medium speed film with general purpose grain, that should be alright for shooting in moderate sunlight at 1/200th of a second. In this case the number is a reasonably direct measure of sensitivity - the higher the number, the more sensitive the film, and the faster you can shoot whilst still getting a similar-looking (though somewhat "rougher", grainier, lower resolution) picture out the other end when processed under the same conditions (actual photo film can be "pushed" or "pulled" somewhat during processing to compensate for under or overexposure, during the secondary part where the developed negative is itself exposed onto the final print paper, rather like adjusting the RAW version of a digital photo in a retouching program - but in both cases there is a limit at each end where the details disappear into the shadows or highlights and can't be recovered, and that's essentially what's governed by the rating). Double the ISO, double the sensitivity, half the exposure time (or aperture area?) for a similarly "lit" image, though you probably lose half of the effective resolution (as in, 1/sqrt(2) in each direction, or maybe 1/cube root(2) in each spatial dimension, and also in the dynamic range one too).

    So, for shooting mostly well-lit scenes with a relatively slow shutter and wide aperture, but capturing a lot of fine detail in the pictures you take, you'd want ISO 100 film, maybe even lower (professional portrait photographers probably use ISO 80 or less, and relatively large-format film instead of plain 35mm?). For taking fast-action sports shots on an overcast autumn afternoon in an urban area, with not very good floodlights on the field, or a lot of indoor/night time shots with a weedy point-and-shoot flash, ISO 400 or possibly higher (I think most single-use cameras are loaded with 35mm ISO-400, and a lot of smaller-frame formats like APS and 110 cartridge didn't tend to offer anything lower than 200 because their limited light capture area meant sensitivity needed prioritising over resolution, even though that was made even worse by the dimensions). Paparazzi trying to get some kind of printable image of a celebrity, snatched in a sudden moment as they hurried past, presumably used ISO 800 or more with their chunky lenses and huge, extra-bright flashguns, so they could still use fast exposure even at night.

    These types of ratings have carried through into digital photography. I don't know if there's actually any kind of direct way of relating the original International Standards Organisation grades (which presumably relied on calibration with a series of lights of different intensities illuminating a particular test image, and a camera with a fixed aperture and exposure...) into digital camera performance, but the same general concepts persist at least - ISO 200 is the "medium", decently-lit daytime setting, which might show a little grain but usually you wouldn't notice. ISO 100 is lower sensitivity but smoother, for when good lighting is guaranteed, and sometimes you might even get 80 or 64 with some models, which more or less eliminates grain entirely. ISO 400 is a more general purpose cloudy/bright-indoor setting, with some slightly noticeable grain but nothing too bad, and ISO 800 is often the highest a lot of cheap cameras will rate, allowing some modicum of indoor night-time (or streetlight) shooting at slightly extended exposure times without needing a flash, but with pretty obvious grain. Again, better cameras offer 1600, 3200 or more, through a combination of better quality sensors (reducing grain at all ISO levels without sacrificing sensitivity - much like the expensive film did) and simply pushing the post-processing even harder, sometimes colouring in the numbers on the display as a warning to the user that the result might look rather noisy, as the CCD signal level isn't riding very much above background voltage any more.

    My phone camera can do 100 to 800, which is OK I guess. But my old Canon Powershot rangefinder digital stills camera (a category whose demise I mourn) offers 80 to 1600, a 20x range instead of 8x, and although I don't often find much use for the lower end of the range (it does allow a slightly longer exposure at full aperture without blowing out the details, but I'm hard pressed to see a difference in detail), that extra bit of sensitivity at the top end, combined with the better lens and sensor too (and the optical stabiliser system) mean it's a much better machine for taking tricky low-light photos. Unfortunately, neither of them give an option for controlling sensitivity during videos.

    Thus we see that actually the exposure control for a camera is threefold, and can often be a compromise, between the varying demands of shutter speed, aperture width, and sensitivity setting (which digital systems bless us with the ability to alter on-the-fly, instead of being beholden to a single choice for an entire roll of film). Depending on what you're shooting, where and when, and how you want to present it, you may want to tweak one or more of those and then have to compensate with the other(s). Happily, the typical control modes on a modern camera allow for a reasonable amount of control over them... the names sometimes vary but the ones of interest here typically comprise:

    AUTO mode (marked as such, or maybe A) - the camera assumes you want a moderate brightness level across the whole frame, under the normal range of lighting conditions encountered everyday (with very dark or very bright seeing it gradually bias towards under- or over-exposure), and with a moderate shutter speed (typically in the 1/60 to 1/500 range) and a medium ISO, plus as wide an aperture as possible unless zoom/focal distance demands otherwise, and will vary all three to try and achieve this without any of them going wildly off into the woods. Different models seem to prioritise particular things (I've one in particular that is convinced that it's perfectly alright to randomly set really long exposures, down to 1/2 second, before it starts to crank up the ISO...) but in general there should be a tendency towards wider aperture, longer exposure and higher ISO in dark places and the opposite in bright ones, with it automatically adapting between shots or through the run of a video. Auto mode is usually deliberately quite locked down to avoid flooding the newbie with too much confusing info, or to give you a "safe" mode to hand it over to a kid with, so the amount of stuff that you can change is usually very limited, sometimes nothing more than changing the flash between "auto" or "off" (but not "forced on"), setting a self-timer and adjusting resolution.

    PROGRAM mode (or just "P", sometimes confusingly "S") - a slightly more flexible setting and generally what I use most. Aperture and exposure are still automatically controlled within similar limits, but you have more control over various items. Particularly, ISO can be changed (either "auto" - which usually limits the maximum to 400 or so - "auto hi" that extends it further and may limit the minimum, or locked to a particular value) so you can balance the overall feel of the pic and what range of speed/width is likely to be used (the name I expect comes from the old auto-program film cameras that detected the speed rating encoded on the body of a 35mm cartridge with a pattern of conductive and insulating squares and adjusted their operating regieme to suit), but also you have greater control of flash (forced-on, slow sync, and flash strength often available), exposure adjustment to compensate for when the automatic setting is either temporarily or consistently lighter or darker than you'd like (usually either one-half or one-third "EV" stops, with a full EV stop being a doubling or halving of brightness and the range usually +/- 1 or 2 stops from full-auto - it's a somewhat crude thing, but still very useful in extreme conditions, especially where some particular unimportant part of the image is "poisoning" the light meter reading to be too low, or more commonly too high, causing the image to wash out or be lost in shadow... eg a lightbulb in the background, low sun at dawn/dusk... it's particularly good for helping to calibrate dashcams, oddly), colour profiles (usually with an option for manual saturation and contrast too), JPG compression strength, manual/auto/macro focus, focus and exposure metering method, that sort of thing. Essentially what you could do with a midrange / prosumer 35mm pocket or superzoom camera around the turn of the century, with some extra digital bells on.

    APERTURE PRIORITY (or just "A", "As", "Ap" or "F") - like Program, but you can alter the Aperture setting, locking it to a certain value in order to achieve a particular effect, with the other settings (ie shutter speed, and - if not itself locked down - ISO) having to compensate for conditions entirely by themselves. Can't say I use it much myself, but then I don't have a really nice camera where you could use it to get some really good depth of field effects, or really crisp macro shots without having to employ extra long exposure and huge ISO. Note that it can't violate any mechanical requirements and if used with a zoom lens might be mechanically forced upwards (or downwards, depending where you start), and then not adjust back down afterwards without manual intervention.
    One particular feature is that, similar to extra high ISO, the camera usually won't go anywhere near its tightest aperture settings unless it really really has to, so if you twiddle the control in that direction you might see some pretty high numbers that you would never have otherwise seen, eg f/11 and beyond on my Canon (usually it doesn't go much beyond 6)...

    SHUTTER PRIORITY (usually "T", "Tp", "Ts"... for "time", I guess?) - as above, but for shutter speed instead of aperture. Lock the exposure time, and the aperture (and optionally ISO) vary automatically to keep the output usable. The adjustment tends to get even further than aperture, when compared to Auto or Program, in that you can only unlock the very fastest or very slowest shutter settings in either Tp or fully Manual mode, otherwise it tends to stay within a more conservative range that can lead to pictures that don't come out at all even if on a tripod (in low light), or more rarely get completely washed out as it won't snap fast enough. Normally the Canon only allows 2s thru maybe 1/2000th (and will step up the ISO as things get darker, through the 1/100th to 1/15th range to try and keep the speed up), and the cheaper backup Nikon won't go slower than 0.5s, but in Ts and Manual it can do as long as 60s or as short as 1/10,000th (or less, even?), and even the Nikon does 15s to 1/8000... Which allows you to go right from taking starry sky pictures outside to getting up close and capturing the smallish writing on the outside of a lit 100w bulb indoors.

    FULL MANUAL CONTROL (typically "M") - As expected, you take full control of aperture AND exposure (with automatic or fixed ISO), in order to really fine tune a shot to your liking. Sometimes EV is still available too, in case you want to bias the ISO setting somewhat without locking it down, or to adjust the range of the subtle "too dark/too bright" warnings that the light meter system provides in this mode and Ap/Tp (it won't stop you taking a picture, but it will e.g. turn the speed/aperture display lettering red or yellow if it thinks there's not enough light coming in to produce a decent average-brightness image... of course, you might WANT one that's extra bright or extra dark).

    Mind that whilst the other modes could be quite useful for various reasons, I'm not really sure if this is particularly useful for a video, to be honest, as the potential for large changes in brightness could lead to your recording going from very bright to very dark with few identifiable details in either case, unless it's got a genuine and effective HDR mode that can be turned on independently of everything else; often I emulate it a little in the standalone cameras by turning the contrast in their colour settings to minimum, and the saturation down halfway, which gives a much more natural look to everything (essentially that's more or less what the camera sees, unfiltered, and the "normal" setting has both of them slightly overcranked by different degrees to make things look more vibrant at the expense of losing some very bright/dark detail and giving a slightly hyperreal cast to everything), but I still wouldn't expect them to be able to cover the usual range of brightness you could encounter just on a 360 degree spin unless I got the settings just-so before starting the run, and the brightest/darkest images would still be hard to properly make out. Really, it's much better for single photos that you want to achieve a particular effect with, or short stylised video clips either in a studio or a single fixed outdoor angle. For everything else, much better to let the camera vary things a bit more if you're not able to rapidly crank it up and down yourself whilst keeping an eye on the live display.
    (that sort of thing is why the upper end prosumer cameras have multiple dials on them still, even with touchscreens etc, so you can rapidly dial all six of the main controls - exposure, aperture, EV, ISO, zoom and focus - up and down using muscle memory and haptic feedback, without having to hunt through various menus and break concentration to stab away at the screen, making the aim wobble)

    On top of which, for video, there's resolution and frame rate. Hopefully you're more familiar with that....

    Resolution, really you want to set 720p as a minimum these days, 1080p if the device and memory card have the speed, capacity, and battery life to deal with it. Going higher is your own choice, and not universally useful at this point. Though, as with still photos, the higher the original resolution, the more you can crop down and digital-zoom without things becoming ugly and pixelated. If you really, really have to set SD, see if a 50 or 60fps progressive mode is available. Avoid interlaced recording like the plague, it's a 20th century relic, and the fact it got OKayed for use with 1080-line HD shows that there are some sick minds in the world. Use 25/30fps instead of it unless you have some properly fast action to capture and need the additional smoothness - the only reason to record 50i/60i any more is if no other high speed mode is available...

    960p, or 1440x1080 non-anamorphic tend to be 4:3 modes (or what gets called "tall" these days!). If you need the extra frame space, then by all means give it a go, but be aware you'll have to either trim it down or pillarbox it for fullscreen viewing on any modern display. It can be useful to allow a roving window within the larger image area (eg 1280x720 within 1440x1080), especially with a fisheye lens, but that's a more advanced processing technique. (1440x1080p anamorphic is still 16:9, just with fatter pixels; "1440p" is something else entirely, basically twice 720p - aka "2.5k" - and not really a standard).

    Unless it's unavoidable, or you really prefer the low-framerate effect for some reason, I'd say it's currently best to record in 60fps progressive*. You can convert down from it with fairly good results, upconversion is a reasonably mature technology (and, honestly, if you can detect the artefacts at 120+ fps, as well as not feeling slightly nauseous, you're some kind of mutant), and if we ignore the remaining stock of PAL and PC CRTs it's the best match for the native scanrate of the vast majority of modern screens, which all default to 60hz.
    (of course, by this we mean 59.94fps, but that 0.06 discrepancy means a single frame jitter every 16.7 seconds even if "60" means 60.00, so it's not a major problem; most videos that have any kind of editing rather than being straight end-trim-and-upload jobs don't have cuts that last longer than 10-15 seconds, so you can engineer it to avoid skips)

    *(or, if your target display is something receiving a PAL broadcast, 50fps... if there's something that's glaringly obvious and annoyingly ugly, it's content that was originally filmed at 30 or 60fps either because it originates from an NTSC area or the creator was ignorant and/or careless, being shown as part of a 50fps PAL TV stream, after poor quality or basically nonexistent framerate conversion. Herky jerky city. I've even seen this crime committed by brands trying to promote the motion smoothness of their new 200hz TV, or the fluidity of a videogame engine... some people just need slapping)

    30fps, or as previously mentioned 20, 15 or even 12 (and the PAL equivalents down to 12.5) can be acceptable substitute rates for reasons of storage space, extra low light levels, or special "8mm" or other timelapse stylised effects, but you wouldn't want to watch a half-hour film at anything below 24 (or half hour TV show at below 25 or 30). There's a reason these standards became established, and it's because anything slower and more jerksome is just plain uncomfortable. Animation can often get away with 12-15ish fps for most of the action, but only for the characters, with a considerable amount of speedlining, fake motion blur and other "hinting" distortions (and the panning across scenes and backgrounds are still at full framerate) to trick the eye into seeing more actual movement than is present.

    At the other end you can go up to 100/120fps or more, but unless you're going to very definitely show it on a lot of HFR displays and you can demonstrate any kind of perceivable difference in a scientifically valid double-blind study, I'd do that only for the possibility of smooth slowmo (replay at 25-30fps = 1/4 speed...) or selectively adding vector-interpolation-based motion blur to a sequence initially shot with a high enough shutter speed to render everything on screen in pin-sharpness. Ditto 200+fps but even more so, as we're decidedly into the realm of just-in-case slowmo capture, and decimating from that speed back down to 60fps without any kind of adaptive blurring may actually toss away a lot of useful motion information that would otherwise have been blurred into the one frame. Additionally, these higher rates obviously burn through a lot more storage space and battery power (and processing power later on), and can limit you to lower resolutions, so it's yet another balance to be made. Really, 1080p at 60fps gives you plenty of flexibility, and anyone who complains about the jerkiness or blockiness at this point needs to be mercilessly robbed^H^H^H^H^H^H befriended in the strongest possible way as their house must be full of all kinds of amazing stuff. If you think you might need more frames or more pixels for cropping or slowmo effects, go up to e.g. 2.5k/1440p (2560x1440) or similar, and/or 120fps, but try not to be forced down to less than 720p or 30fps as a compromise. And check that your card is fast enough to withstand it before recording anything important!

    Anyway, hopefully that should be some kind of useful rundown. Have fun and happy experimenting, and please go make some more actually interesting and nice-to-look at videos, because there's too much crap on youtube at the moment


    NB one thing I didn't mention above as it's only really relevant to stills cameras, and then typically SLRs or really good compacts (BUT these do get used as video cameras quite a lot, too), is the realm of physical filters; either ones that attach externally onto the lens or the camera body, or sometimes built into the camera itself. Naturally these almost always reduce the light level in different ways, but it can be worth it for a particular effect. External ones come in all kinds of flavours - colour tinting, either overt shades or subtle fill-ins to make overcast skies look more blue or add a fake "sunset" cast to things, micro-etched jobs that add starburst or other effects to the "bokeh" (spots of light that are defocussed into circles due to DoF effects), polarisers that help reduce glare and enhance natural contrast especially where low sunlight or reflections from water are involved, and "NDFs" - neutral density filters that universally reduce the amount of light coming in without tinting or otherwise affecting it, so you can use a longer exposure and/or wider aperture to add deliberate motion-blur and soft-focus effects without overexposure, even though you can't drop the ISO any further. Usually they're rated in terms of how many stops they reduce the scene brightness by, eg a "3 stop NDF", so with otherwise identical settings you can open the aperture up that much or slow the exposure accordingly without changing the brightness of the scene. It can be a particularly popular option for natural scenes or crowds, roads, etc where you want a drawn-out timelapse blur effect (eg that particular one of a stream or waterfall looking like a carpet of gossamer silk is achieved this way, and I've actually managed to produce similar with a compact camera in the height of summer by clamping the lenses of two different pairs of cheap sunglasses to the front with duct tape) during the daytime.

    Also that of bracketing, which is both a method of taking multiple shots in an uncertain lighting condition and picking the best one later, and of combining those shots to produce an HDR image; the camera either automatically, or under the control of a VERY steady-handed user, takes two, three, or more pictures in as rapid succession as possible - one (or more) with a negative EV, optionally one neutral (ie EV 0) and one (or more) positive, to create a series of otherwise near-identical images that have known differences in exposure/brightness between them, and a maximum and minimum value therefore rather extended from the expected norm of EV 0, which allows the camera or a suitable program to align the multiple pictures, adjust their contrast and brightness accordingly so the midtones line up and the very darkest colour = black, very brightest = white, then paint in any parts which clipped to flat black in the darker pictures with detail from the brighter ones, and vice-versa for what clipped to white in the brightest ones, possibly with some local tweaking of levels in different parts of the image to improve overall contrast (else it may just look washed out - though going too far with that effect leads to the weird "oversharpened" or "haloed" look that HDR images can have), to create a final image with a lot more brightness range than would otherwise be possible with the typical dynamic sampling width of a typical CCD. Given that there's a particular difference in energy between the brightest and darkest at a particular ISO (as it's basically based on amplifying the semi-analogue signal by a certain amount...), the more pictures there are, the wider the range can be, and the greater the overlap they can have which improves the quality as well. The most I've usually seen is 5 steps, 1EV apart, but you can have 3 or even just 2 that are each 2EV apart (-2, 0, +2, or just -1 and +1) without it looking too awful. -2 and +2 would probably not work
    (that last one is definitely not practical for video unless you manage to hack a 3D camera to have different exposures from each of two very closely spaced lenses, or have something running at 240fps without much motion of its own, taking a series of 4 shots every 1/60th sec at +1.5, +0.5, -0.5 and -1.5EV...)
    -= She sez there's ants in the carpet, dirty little monsters! =-
    Back after a long time away, mainly because I now need to start making up vidcapped DVDRs for work and I haven't a clue where to start any more!
    Quote Quote  
  12. Far too goddamn old now EddyH's Avatar
    Join Date
    Jan 2003
    Location
    Soul sucking suburbia! But a different part since I last logged on.
    Search Comp PM
    Click image for larger version

Name:	EcrTuZF.jpg
Views:	90
Size:	43.5 KB
ID:	41494

    And in a stunning demonstration of the old adage re words, pictures, and the relative worth of each, I found the image I was thinking of. It is, however, 43.4KB, so technically that's 22,255 words, or 22,528 if we count the wasted disc space as well. Hm.

    Note how, although one does not absolutely imply the other, you can see a relationship between each line as well - if you move to the right along the top one, you'll probably have to move to the right along at least one of the others in order to keep the same exposure level (however, the relationship between exposure and ISO is mirrored - if you don't change the aperture, but move left along the exposure line, you'll have to move to the right on the ISO line). For some reason it hasn't included the effects on brightness of changing the shutter speed or the aperture, but just take it that going right along the top line should see the picture become darker, and on the middle line it should become lighter.

    And actually they've ballsed the ISO one up. The picture should become noisier, but the background should have started out grey in the ISO50 picture and got gradually lighter throughout.
    What I'm saying is it's imperfect, but it should give you a visual idea of what I wrote.

    Oh and remember if you shoot at 120fps you obviously aren't going to be able to use a 1/60th exposure, but you *might* be able to blend frames together with an additive filter to create a strange temporal overlap effect where each resulting frame holds information from 1/60th of a second but only displays for 1/120th, sort of like a "sliding window" that's centred on a certain point and shows what was happening upto 1/120th of a second before and after it. Or two 1/60th sec videos, with 1/120th sec exposure, which were started almost in synch but 1/120th of a second apart, being projected onto the same part of a wall with their original temporal relationship maintained.

    Yeah. Something like that. Erm.
    (this effect is hella ugly when done at lower framerates - a good example being a particularly annoying fabric softener advert that's doing the rounds at the moment with some animated characters formed of clothes standing around a laundry basket, which seems to have been animated at maybe 10fps then smoothed by partially overlapping the frames - but it might come good if the speed is increased)


    PS I have no idea about the framedropping schemes used in different cameras, that's something that DEFINITELY has to rely on trial, error, and poking at the files for yourself. One would have hoped that the rate would be derived from a common clock, e.g. 119.88fps or some higher multiple, and all it would do to run at 30fps was divide that clock by 4, or divide by 5 to run at 24fps. (Or, ultimately, a 600khz clock with a choice of 1/1000 and 1/1001 dividers, so the output from that can be further sliced by 5, 6, 10, 12, 20, 24 or 25 to provide 120 (or 119.88), 100, 60 (or 59.94), 50, 30 (or 29.97), 25 and 24 (or 23.976) output rates... possibly even one that's faster still and is also subdivided for the line and pixel rates on the way to also producing the frame rate.

    Kinda disappointing that it basically still captures at the same 30fps rate and simply neglects to save every 5th frame in order to have 6 fewer frames per second to encode onto the memory card (maybe in exchange for having a slightly higher resolution? did you say it was 900p instead of 720p, or something of that nature? If not then it's *definitely* stupid, because it offers no advantage at all other than maybe 25% more recording time at the same compression level - it won't be capturing any extra light, because each recorded timeslice is still 1/30th sec, and in fact it's throwing 20% of it away).

    That's actually something I'd go bitch at the company themselves about, because it's a total bodge that shouldn't be allowed to stand. The only saving grace might be that it really IS recording at a proper 24fps (and so getting 25% more light, potentially?), but there's some issue with the file format or chipset that means it can't *save* it as that, instead having to shoehorn it into a rigid 30fps container, maybe with a buffer (fairly necessary for MPG encoding anyway) holding at least 4 frames that can take them in at 24fps and spool them back out at 30 into the encoder (if it can't itself be slowed down), with some extra code to issue a flag for the 5th frame, instead of any actual data, that forces the writing of a "skip" or "repeat frame" placeholder tag in that place in the file. Sort of like how progressive 24-frame works on DVDs IIRC. In other words, a bit of a lashup, and probably something our players are misinterpreting.

    The acid test is, if you take that source and process it such to first a/ just speed it up to 30fps (including the sound), and then for a second copy b/ insert a duplicate copy of the one of the frames either side of each "jump" (or a blended average of the two) in-between them and output again at 30fps (not changing the sound!), which one of them then looks smoother.

    If it's the first one, then it's actually just an artefact of the display you're watching it on, which if it's 60hz necessarily has to show "2.5" native frames for each frame in the video and so will have quite a bit of jitter (at best, showing every other frame for 3 vsyncs, vs 2 vsyncs for the other... too-slow, too-fast, too-slow, too-fast... and the division might not be as clean as that if it's showing a 23.976fps video on a 60.0x hz screen, or one that uses a higher (or lower! LCDs that run at 40hz in power-saving mode are a thing in some laptops now) refresh rate)... and if the player software is a bit rubbish it might even only cycle at 30fps, so it literally can't show 24fps content other than by doubling every 4th frame (as 30 = 24 x 1.25). Which means the content is still valid 24fps material.

    If it's the second, then sadly the original suspicion was correct, and it's recorded 4 frames at 30fps, then paused for 1/30th of a second before carrying on. A custom motion-detection framerate smoothing script might be able to do something about it, either just patching between the two frames either side of the gap (=30fps with a completely new frame inserted to disguise the jump entirely), or trying to massage their times closer together by applying some adaption to the existing frames either side of the jump based on differences between them and the two that are little further away in time (=24fps output, with 2 frames out of every 4 having some morphing to incorporate a little more of their neighbour on one side and less of the other, and so be slightly displaced in the original timeline compared to where they previously claimed to be). But that's a bit of a faff, isn't it.

    If they both look a bit shoddy, still, it's possible that it has actually recorded at a proper 24fps but has inserted a blank frame to pad out to 30... there's probably a fairly simple script (use of "select from"? or is that php_mysql invading my dreams again?) that can nix the surplus frame in that case, and you don't even have to mess with the sound if you just set the resulting video stream to play at the correct speed.

    Let's just hope they don't do something as stupid as that with 25 or 50fps recordings. Or indeed that the 30fps ones aren't actually capturing at 60fps but discarding every other frame...

    .....ok i'm done now
    -= She sez there's ants in the carpet, dirty little monsters! =-
    Back after a long time away, mainly because I now need to start making up vidcapped DVDRs for work and I haven't a clue where to start any more!
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!