Color correction - Where to begin ?

[JDługosz]

I've not needed it before, but the lighting must have been funny — different on subject than background.
I tried the various color filters that came with AVIDEMUX, but can't figure out what to do with the sliders to make it look right. I looked for tips online and it shows displays like a radar screen and histograms, that are used by pros.

I have Sony (Vegas) Movie Studio Platinum 12, which mentions a much more primitive color correcting display in the help file, but for some reason no extensions/plug-ins are loaded even though I can see C:\Program Files\@AV\MovieStudio\Video Plug-Ins\colorcorrector.dll exists, along with others.

I wonder if there is a front-end that uses some display to help me choose settings, that I can then feed to one of the free filters on AVIDEMUX (my preferred program to simply trim and encode for You-Tube)? Or, some other way to figure it out.

[Baldrick]

I removed the special character(—) from the thread title. It made the thread non clickable.

[sanlyn]

You're not giving us enough info concerning the details of the video you're working with. There are lots of "videos", in many different formats, using many different encoders. Ultimately you'll get the best advice if you provide a sample of the original, unprocessed source video. If you don't know how to make and upload a short sample, just ask.

To begin, we can get information about your source video easily if you download the free MediaInfo utility and open your video with it. Use the "Tree" view in the MediaInfo window, copy the text, and paste it into a reply here. You can also export MediaInfo data as a text file. NOTE: When installing MediaInfo, ignore or reject any "special offers" to install additional software, toolbars, etc. The advertising helps to keep MediaInfo cost-free for its users, but you won't need the extra software.

We can't accomplish very much with a still image But at least we can see from your post that there are several problems. Your video (or your image, at least), has invalid luma and chroma levels (crushed darks, blown-out highlights, oversaturated reds, etc.). You can't correct those problems with the primitive image controls in the software you mention -- and even if you could, your software uses multiple re-encodes from RGB and lowers the quality of your source. If you were doing simple edits with no major corrections, that wouldn't be too terrible, but the corrections you need to make will lower video quality with the software you are using.

Try using MediaInfo to give us more information. If you need help using it, let us know.

[budwzr]

The filters in Movie Studio are on the timeline.



There are three domains, Track, Event, and Global. The Global one is on the Video Buss Track.

The color tools are too many to discuss all of them, but you have everything you need to put out professional results.

[sanlyn]

I know. They won't be much help.

[JDługosz]

You're not giving us enough info concerning the details of the video you're working with. There are lots of "videos", in many different formats, using many different encoders. Ultimately you'll get the best advice if you provide a sample of the original, unprocessed source video. If you don't know how to make and upload a short sample, just ask.

I did not suppose that the exact encoding used by my camera mattered to finding a color-correcting tool, so long as the program could in fact load it. Why does it matter? Here is the report from MediaInfo:

Format                                   : AVC
Format/Info                              : Advanced Video Codec
Format profile                           : Baseline@L5.0
Format settings, CABAC                   : No
Format settings, ReFrames                : 1 frame
Format settings, GOP                     : M=1, N=12
Codec ID                                 : avc1
Codec ID/Info                            : Advanced Video Coding
Duration                                 : 3mn 15s
Bit rate                                 : 32.4 Mbps
Width                                    : 1 920 pixels
Height                                   : 1 080 pixels
Original height                          : 1 088 pixels
Display aspect ratio                     : 16:9
Original display aspect ratio            : 16:9
Frame rate mode                          : Constant
Frame rate                               : 23.976 fps
Color space                              : YUV
Chroma subsampling                       : 4:2:0
Bit depth                                : 8 bits
Scan type                                : Progressive
Bits/(Pixel*Frame)                       : 0.651
Stream size                              : 755 MiB (95%)
Language                                 : English
Encoded date                             : UTC 2014-02-01 20:13:38
Tagged date                              : UTC 2014-02-01 20:13:38
Color primaries                          : BT.709
Transfer characteristics                 : BT.709
Matrix coefficients                      : BT.601

We can't accomplish very much with a still image But at least we can see from your post that there are several problems. Your video (or your image, at least), has invalid luma and chroma levels (crushed darks, blown-out highlights, oversaturated reds, etc.).

You can't correct those problems with the primitive image controls in the software you mention -- and even if you could, your software uses multiple re-encodes from RGB and lowers the quality of your source. If you were doing simple edits with no major corrections, that wouldn't be too terrible, but the corrections you need to make will lower video quality with the software you are using.

Worse than I thought, then.

I've attached two sample clips, showing the zoomed-out and the close-up.

[JDługosz]

I know. They won't be much help.

Avisynth and VirtualDub. Still doesn't look right. We need the original video before it got crushed and clipped in Movie Studio.


[Attachment 23457 - Click to enlarge]

That still is from the camera's original file. I've not done anything with it other than take a screen-shot of the Preview image being displayed.

[poisondeathray]

The color tools are too many to discuss all of them, but you have everything you need to put out professional results.

The studio version of vegas doesn't have scopes, you need the pro version . You basically do it by "eyeballing" it - not very accurate


I bet very likely this was shot on a canon or nikon DLSR (typical moire patterns , soft mushy picture) , 601 matrix

That still is from the camera's original file. I've not done anything with it other than take a screen-shot of the Preview image being displayed.

That's the problem - there is data in YUV (the original video colorspace), that is not represented in a RGB (different color model or color space) screenshot . There is not a direct 1:1 mapping between YUV and RGB (8bit YUV is much larger than 8bit RGB in terms of gamut), and using a standard "rec" matrix will clip superbrights and darks

Anyways, I'm sure sanlyn will walk you through it

[JDługosz]

The filters in Movie Studio are on the timeline.

Thanks! If only the "Help" page had that little arrow. It says "The Color Corrector plug-in helps you adjust colors in your video by independently adjusting the color of low, middle, and high color tones." and points out "The Color Corrector plug-in is available only in Vegas Pro and Vegas Movie Studio Platinum software" and shows a picture of part of the control, but nothing tells you that Color Correction is found in "Video FX". I was expecting a desktop window layout preset for color work, or color listed specifically on a menu somewhere, and all the "plug-in" or "extension" submenus were blank.

[sanlyn]

Thank you for the samples.

Another reason your image is difficult to work with is that it's not 'from the video", it's from your computer display. But thanks again. Will take a look.

There is only 1 frame in your sample1.mkv

[sanlyn]

If only the "Help" page had that little arrow. It says "The Color Corrector plug-in helps you adjust colors in your video by independently adjusting the color of low, middle, and high color tones." and points out "The Color Corrector plug-in is available only in Vegas Pro and Vegas Movie Studio Platinum software" and shows a picture of part of the control, but nothing tells you that Color Correction is found in "Video FX". I was expecting a desktop window layout preset for color work, or color listed specifically on a menu somewhere, and all the "plug-in" or "extension" submenus were blank.

Even in the pro version, you'll have to make at least basic corrections in the original colorspace (YUV) before moving to RGB. I believe Vegas Pro can work in YUV (correct me on that, folks).

Neither Vegas Pro nor Movie Studio is a smart rendering editor with AVCHD. If you made this cut in Movie Maker, it has likely been converted to RGB, back to YV12, and re-encoded. Be that as it may, the color in the mkv isn't great, but it's much better than in your image.

Anyway, still looking....

[budwzr]

Vegas Movie Studio[/url] Platinum software" and shows a picture of part of the control, but nothing tells you that Color Correction is found in "Video FX". I was expecting a desktop window layout preset for color work, or color listed specifically on a menu somewhere, and all the "plug-in" or "extension" submenus were blank.

You can make any layout you want, and there are a couple presets.

[JDługosz]

The color tools are too many to discuss all of them, but you have everything you need to put out professional results.

The studio version of vegas doesn't have scopes, you need the pro version . You basically do it by "eyeballing" it - not very accurate[/QUOTE]

Yea. It was a good deal at the time, getting just the editing capability that I needed at a reasonable price. It's rather good except where it just falls flat.

I bet very likely this was shot on a canon or nikon DLSR (typical moire patterns , soft mushy picture) , 601 matrix

e.g.
http://colorbyjorg.wordpress.com/2011/01/14/canon-dslr-video-uses-bt-601-sd-matrix-instead-of-bt-709-hd/

In contradiction to what you might assume, the Canon DSLR cameras video conforms to the ITU-R BT.601 standard for SD video recommendations, instead of the ITU-R BT.709 HD standards. This means that several suppliers of software for video- and film-makes try to convert the color decoding matrix of the Canon DSLR video footage. An example is the new 5DtoRGB software that assumes (in it’s [sic] helpfile) that the Canon 5D (1D/7d/550d/T1…) uses the ITU-R BT.601 color decoding matrix (“Primary chromaticities”). THIS IS NOT THE CASE!

I've not looked into color space mapping, as I normally just trim and encode with x264. The metadata that says what it is must be doing the job, because it looks "fine" played with VLC or on You-Tube. But that could be why certain "other" players don't give the same colors.

But, I suppose for color correcting is it important that the correcting tool understand what space it's in, or the sliders won't give the expected results. So, what should I do (keeping with shooting video with my Canon 60D and Magic Lantern)? I don't mind huge intermediate files.

That's the problem - there is data in YUV (the original video colorspace), that is not represented in a RGB (different color model or color space) screenshot . There is not a direct 1:1 mapping between YUV and RGB (8bit YUV is much larger than 8bit RGB in terms of gamut), and using a standard "rec" matrix will clip superbrights and darks

Ah, good to know. So even a single frame from video file is more accurate and doesn't add to the problems with its own color space conversions. But, being sRGB, the Canon video ought to do better in converting to RGB for a web browser. But that's at the mercy of the previewer and the screen grab won't get the ICC profile... I see your point.

[JDługosz]

You can make any layout you want, and there are a couple presets.

That's only on the "pro". I'm missing that preset, and there is no Video Scopes view.

http://www.sonycreativesoftware.com/vegaspro/compare

I have the second column, the $80 version. For $320 more you get scopes, 24-bit sound, and more device captures.

[JDługosz]

If you made this cut in Movie Maker, it has likely been converted to RGB, back to YV12, and re-encoded. Be that as it may, the color in the mkv isn't great, but it's much better than in your image.

I made the samples using Avidemux 2.6.1, 64 bit, on Windows. I set video to "Copy" so it did not re-encode anything at all; just copied to a new container file. The camera writes MOV files. You can see that it's compressed as simply individual keyframes but with 38mb/s.

Anyway, still looking....

Thanks.

[budwzr]

Oh really? I thought it had more features.

Anyway, the banding artifacts and blown highlights are a big problem, even if you have fancy tools. Hopefully Sanlyn can advise more on how to fix that.

[JDługosz]

Oh really? I thought it had more features.

Anyway, the banding artifacts and blown highlights are a big problem, even if you have fancy tools. Hopefully Sanlyn can advise more on how to fix that.

It usually works better. So I blame the lights.

[smrpix]

It usually works better. So I blame the lights.

the lighting must have been funny — different on subject than background.

The video from a technical perspective is slightly hot and a little too red, but not bad and perfectly plausible to someone who wasn't there. Your references to the lighting make me think there's something else you're expecting which is more an aesthetic issue than a technical one. Can you describe what you think is wrong?

[poisondeathray]

I've not looked into color space mapping, as I normally just trim and encode with x264. The metadata that says what it is must be doing the job, because it looks "fine" played with VLC or on You-Tube. But that could be why certain "other" players don't give the same colors

Yes, I only mentioned it because all those clues strongly suggest this is Canon DLSR footage

Those VUI flags are embedded by the recording module, not the sensor or camera processing (ie. they are just flags after the fact and don't necessarily reflect what is happening in processing after debayering). Most consumer level Canon and Nikon's share the same recording module (and they record using crappy baseline AVC, very mushy picture), and they all tag it the same way. Those VUI flags are just suggestions as to how to view the YUV data in RGB (ie. suggestions to the player how to convert to RGB for display) . Most players actually ignore those flags, except if you have it specially configured with madVR. So if you feel those colors look basically ok in your screenshot (other than the problems like saturation and levels), it's actually using 709 in your screenshot (ie. if you took screenshot with VLC, it's using Rec709 for the RGB conversion, not 601) . The difference between 709 and 601 isn't large, the colors will be slightly shifted, that's all. You have the choice to use whatever matrix you want, in programs like avisynth, 5d2rgb

Important thing you should be aware of - the choice decoder plays a large role in what levels you get for the native MOV quicktime files. For example , if the software you use uses quicktime the levels will be clamped to 16-235. Most open source /libav software will decode full range 0-255 . So if you open it up in avidemux, which is libav based, the levels will be full range - that's why your screenshot looks clipped. If you took the screnshoot out of quicktime, it wouldn't be as clipped (because the level in YUV are clamped). Old versions of vegas used to use quicktime for MOV decoding, not sure about the newer versions, I think they moved away from quicktime . Quicktime also uses a completely different gamma, so it will look different in QT player regardless of levels

Vegas in 8bit mode will use studio RGB (that's not the same thing as sRGB) - so superbrights/darks will be preserved. Studio RGB is almost the same thing as "full range" or a PC matrix if you're using avisynth. It's tough to do this properly in the studio version because you have no scopes - you're working half blind.

But, I suppose for color correcting is it important that the correcting tool understand what space it's in, or the sliders won't give the expected results. So, what should I do (keeping with shooting video with my Canon 60D and Magic Lantern)? I don't mind huge intermediate files.

Yes . There are multiple ways of doing this - it depends on what software you're comfortable with, how complex you want to get, what sorts of manipluations you want to do

That's the problem - there is data in YUV (the original video colorspace), that is not represented in a RGB (different color model or color space) screenshot . There is not a direct 1:1 mapping between YUV and RGB (8bit YUV is much larger than 8bit RGB in terms of gamut), and using a standard "rec" matrix will clip superbrights and darks

Ah, good to know. So even a single frame from video file is more accurate and doesn't add to the problems with its own color space conversions. But, being sRGB, the Canon video ought to do better in converting to RGB for a web browser. But that's at the mercy of the previewer and the screen grab won't get the ICC profile... I see your point.

Yes, examining the native video file is important .

The video isn't in sRGB, it's stored in Y'CbCr (or YUV) .

Actually the main point is you're throwing away information before even correcting it. If you convert it to RGB using sRGB, Rec601 or Rec709 - those values Y' 0-15, 236-255 , CbCr 16-240 are clipped (clipped, not clamped). Same thing in VLC, because it's libav based.

[budwzr]

What I see is too much EV difference between foreground and background, so in that case you're "hosed from jump".

[jagabo]

A quick Tweak(sat=0.75, cont=0.88, bright=12, coring=false) followed by a rec.601 conversion to RGB doesn't look too bad:



cont=0.86 and bright=15 would be closer to a full PC to rec levels adjustment.

Every editor has simple contrast, brightness, and saturation filters like those. The scales they use may be different. sat=0.75 might be called "sat=75 percent" or "sat=-25 percent", for example.

[Selur]

simply using

ColorYUV(autowhite=true)

the source looks fine too

[jagabo]

I tried to use only the adjustments that are common to virtually all editors. Autowhite won't be available on some editors or may behave differently.

[Selur]

ah, okay, it was just the first thing I thought to try out.

[budwzr]

Yeah, Post#21 looks better, but I see a purple tinge. Anybody else see that too? ( I'm Trying to do a Poorman's Monitor Calibration)

Also, the lighting in the scene is mixed type. Some warm / Some cool. So which ever one is WB'd properly, the other goes purply.

[JDługosz]

(and they record using crappy baseline AVC, very mushy picture),

What do you mean by "mushy"?

to 16-235. Most open source /libav software will decode full range 0-255 . So if you open it up in avidemux, which is libav based, the levels will be full range - that's why your screenshot looks clipped. If you took the screnshoot out of quicktime, it wouldn't be as clipped (because the level in YUV are clamped). Old versions of vegas used to use quicktime for MOV decoding, not sure about the newer versions, I think they moved away from quicktime . Quicktime also uses a completely different gamma, so it will look different in QT player regardless of levels

Vegas in 8bit mode will use studio RGB (that's not the same thing as sRGB) - so superbrights/darks will be preserved. Studio RGB is almost the same thing as "full range" or a PC matrix if you're using avisynth. It's tough to do this properly in the studio version because you have no scopes - you're working half blind.

The original screen shot was from the file browser. Although I'm not using Windows' built-in File Explorer, I'm sure the preview window still just asks the OS to cough it up. Yeah, “Preview created by the shell.”

Yes . There are multiple ways of doing this - it depends on what software you're comfortable with, how complex you want to get, what sorts of manipluations you want to do

I don't mind complex. I'm a software developer by trade. Though I would find having to use three different OS's in the workflow to be awkward.

I'm supposing I should be able to preprocess the camera's files, or use specified settings in the decoder, or attach specific profile data to the original file, or something like that, to generally make it work better.

The video isn't in sRGB, it's stored in Y'CbCr (or YUV) .

I know. I said "converts to sRGB" easily, as the sRGB primaries are what the camera used originally. So doing a conversion to RGB won't have a gamut mismatch from what Windows likes.

[JDługosz]

The video from a technical perspective is slightly hot and a little too red, but not bad and perfectly plausible to someone who wasn't there. Your references to the lighting make me think there's something else you're expecting which is more an aesthetic issue than a technical one. Can you describe what you think is wrong?

His face is orange and lacks tonal range. Good point though, I'm not worrying about the colors of the clothing or floor, just so long as it doesn't look "candy-ish" or desaturated.

[JDługosz]

simply using

ColorYUV(autowhite=true)

the source looks fine too

What is that command for?

[poisondeathray]

What do you mean by "mushy"?

It's codec "mushiness". It's a term people use to describe how the picture is soft & impaired from the poor compression . It's typical of poor AVC compression, and tends to affect shadow areas disproportinately. It's sort of a glazed over, low detail look. It's especially bad for early DSLR's , because they use baseline AVC (no cabac, no b-frames) - the bitrates are decently large 40-50Mb/s that you 'd expect a sharper picture - but still looks not that great for that bitrate - it falls apart when anything taxes the compression . Other cameras at 1/2 that bitrate produce cleaner/sharper pictures (of course there are dozens and dozens of other factors contributing, but a large difference is the compression) .

I don't know if ML raw is fully working for this model yet, but it should significantly improve the results. (Part of the "mush" is actually due the processing and line skipping , but a major part is from the onboard compression). Search for A/B comparisons of ML raw vs. normal compression - you 'll see fine details, almost a grain like pattern with the raw (everything is preserved), but the normal compression discards a lot of the detail giving that "mush" or plastiky no detail look. If you're going the raw route, search the various dlsr forums for workflows. Most people will be using resolve (bit of a learning curve, but there is a bit of reading to do about basic color correction like levels, hue, saturation, color spaces no matter what program or workflow you choose)

This is not the same camera, but it's just a typical comparison of ML raw vs. native h.264 "mush" (and the h.264 isn't that mushy here, it's a lot worse when there is motion)
http://www.eoshd.com/content/10338/low-light-test-of-5d-mark-iii-raw-vs-h-264





And if you're not going the raw route (maybe you have write speed or reliablity concerns), there are lots of free ways to do basic color correction with monitoring in avisynth, vdub, (avidemux doesn't have waveform/histogram etc... monitoring) on the native recordings

What is that command for?

That's avisynth for "auto" white balance.

[sanlyn]

Yeah, Post#21 looks better, but I see a purple tinge. Anybody else see that too?

The tinge isn't purple. It's red. The Autowhite image below it is too green (very dark cyan hair and microphone, among other things).

Still working......