VideoHelp Forum




+ Reply to Thread
Results 1 to 21 of 21
  1. I need help setting the brightness level when capturing an old beta tape from 1985. The more I increase the brightness, the more detail I see, but I also introduce a lot of noise.

    Here's my first capture with low brightness on my capture card:
    Image
    [Attachment 85060 - Click to enlarge]


    Here's my second capture with medium brightness:
    Image
    [Attachment 85061 - Click to enlarge]


    Here's my third capture with higher brightness:
    Image
    [Attachment 85062 - Click to enlarge]


    The first capture seems dark to my eyes, but it also does a good job of hiding those weird black shadows to the right of the people and objects. Those black shadows (?) are especially noticeable to the right of the head of the guitar player on the far left in the second and third captures. Not sure if that's something that will go away when deinterlacing or in post-processing, but it seems like something that would be hard to fix after capture.

    Here's an image slider of brightness low vs high: https://imgsli.com/MzQxMDMy

    Here's another slider of brightness medium vs high: https://imgsli.com/MzQxMDM1

    I kind of lean towards the "medium" capture as the best compromise between leaving in the details and bringing out those weird black shadows (for lack of a better term).

    Thoughts?
    Quote Quote  
  2. Capturing Memories dellsam34's Avatar
    Join Date
    Jan 2016
    Location
    Member Since 2005, Re-joined in 2016
    Search PM
    Even with the highest brightness your capture looks still below the legal limit, You want to capture with a proper level and worry about noise later, You can address the noise in post with some DNR filters. What capture software you're using?
    Quote Quote  
  3. Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]


    The capture is ok and shoudn't turn night/deep black into day/gray.
    You may increase the gamma somewhat in order to stretch the darks a bit, but there is nothing basically wrong with the capture levels IMO.
    If you prefer to tweak it do it in post.
    Last edited by Sharc; 24th Jan 2025 at 07:48.
    Quote Quote  
  4. Member
    Join Date
    Jun 2024
    Location
    Perth, Western Australia
    Search PM
    Most of the frame is blackness, what with the large expanse of black curtain, and the audience in virtual darkness.

    In those days of low res video they had to make the most of what was in that small frame. Especially on a wide shot like this, the viewer craves more detail. If this gig had been set up for video recording they could have made the backdrop a little lighter, with a little pattern or color, and brought the audience lighting up a little, giving viewers more of interest to look at on the wide shot.

    At least the performers wore lighter coloured jackets. Black jackets would have have blended in with the backdrop, giving the performers disembodied heads and hands.

    The camera operator appears to have exposed correctly for the performers and you've probably captured all that's in the frame which is all that we can do.
    Last edited by timtape; 24th Jan 2025 at 20:09.
    Quote Quote  
  5. Originally Posted by dellsam34 View Post
    Even with the highest brightness your capture looks still below the legal limit, You want to capture with a proper level and worry about noise later, You can address the noise in post with some DNR filters. What capture software you're using?
    Thanks dellsam. I'm using VirtualDub to capture. I am going to try another capture with the brightness set a bit higher still and see what I come up with.
    Quote Quote  
  6. Originally Posted by timtape View Post
    Most of the frame is blackness, what with the large expanse of black curtain, and the audience in virtual darkness.

    In those days of low res video they had to make the most of what was in that small frame. Especially on a wide shot like this, the viewer craves more detail. If this gig had been set up for video recording they could have made the backdrop a little lighter, with a little pattern or color, and brought the audience lighting up a little, giving viewers more of interest to look at on the wide shot.

    At least the performers wore lighter coloured jackets. Black jackets would have have blended in with the backdrop, giving the performers disembodied heads and hands.

    The camera operator appears to have exposed correctly for the performers and you've probably captured all that's in the frame which is all that we can do.
    There are many parts of the video where the camera operator zooms in, and the picture is quite good in those parts. I shared the captures of the wide shot just because that's where those weird black shadows to the right of heads and images thing stick out the most. On the zoomed in shots, it's less noticeable.
    Quote Quote  
  7. Originally Posted by Sharc View Post
    Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]


    The capture is ok and shoudn't turn night/deep black into day/gray.
    You may increase the gamma somewhat in order to stretch the darks a bit, but there is nothing basically wrong with the capture levels IMO.
    If you prefer to tweak it do it in post.
    Do you think increasing brightness more will lead to more day/gray? I guess I can give it a shot and report back
    Quote Quote  
  8. Originally Posted by Sharc View Post
    Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]
    Before the video was converted to RGB for the sample images it may have had lots of Y<16 values. On conversion to RGB all those illegal Y values were converted to RGB=0 (you can't have RGB values below 0 in PNG images). On conversion from the images back to YUV all the RGB=0 values were converted to Y=16. There's no way to differentiate those that were originally less than 16 from those that were 16.
    Quote Quote  
  9. Originally Posted by jagabo View Post
    Originally Posted by Sharc View Post
    Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]
    Before the video was converted to RGB for the sample images it may have had lots of Y<16 values. On conversion to RGB all those illegal Y values were converted to RGB=0 (you can't have RGB values below 0 in PNG images). On conversion from the images back to YUV all the RGB=0 values were converted to Y=16. There's no way to differentiate those that were originally less than 16 from those that were 16.
    Oh, fooled by the clipping of the PNG screenshots. Thanks for the hint.
    So the OP may upload a sample of his YUV capture rather than PNG images. If the ultrablacks are not clipped these can usually be set right in post, if the Y is clipped he can try to readjust the "proc-amp" settings (depending at which stage the clipping happened it may help or not).
    Last edited by Sharc; 25th Jan 2025 at 02:26.
    Quote Quote  
  10. Originally Posted by theseeker2 View Post
    Originally Posted by Sharc View Post
    Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]


    The capture is ok and shoudn't turn night/deep black into day/gray.
    You may increase the gamma somewhat in order to stretch the darks a bit, but there is nothing basically wrong with the capture levels IMO.
    If you prefer to tweak it do it in post.
    Do you think increasing brightness more will lead to more day/gray? I guess I can give it a shot and report back
    See jagabos post. Upload a video sample of your YUV capture rather than screenshots for better advice. My bad.
    Quote Quote  
  11. Member
    Join Date
    Dec 2024
    Location
    India
    Search PM
    Medium brightness looks like the best option. It keeps the details clear without too much noise. You can use VirtualDub or similar tools later to fix the black shadows.
    Quote Quote  
  12. Member
    Join Date
    Jun 2024
    Location
    Perth, Western Australia
    Search PM
    Originally Posted by theseeker2 View Post
    Originally Posted by timtape View Post
    Most of the frame is blackness, what with the large expanse of black curtain, and the audience in virtual darkness.

    In those days of low res video they had to make the most of what was in that small frame. Especially on a wide shot like this, the viewer craves more detail. If this gig had been set up for video recording they could have made the backdrop a little lighter, with a little pattern or color, and brought the audience lighting up a little, giving viewers more of interest to look at on the wide shot.

    At least the performers wore lighter coloured jackets. Black jackets would have have blended in with the backdrop, giving the performers disembodied heads and hands.

    The camera operator appears to have exposed correctly for the performers and you've probably captured all that's in the frame which is all that we can do.
    There are many parts of the video where the camera operator zooms in, and the picture is quite good in those parts. I shared the captures of the wide shot just because that's where those weird black shadows to the right of heads and images thing stick out the most. On the zoomed in shots, it's less noticeable.
    Yes not surprised the cameraman zoomed in for better detail on individual musos. Essential. Many unskilled operators would just leave the camera on wide shot, not realizing that facial expressions and the fingerwork of the guitarists would be almost impossible to see.

    The "shadows" are interesting. They match the bright profiles of the musicians directly to their left but interestingly there is less or no electronic noise in them. Not sure what to make of them.

    But as already touched on, this shot illustrates that often we cant separate the wanted picture details from the unwanted noise or artifacts. If we want to see all the wanted low level picture detail we may have to leave the noise in.
    Last edited by timtape; 25th Jan 2025 at 06:56.
    Quote Quote  
  13. Originally Posted by Sharc View Post
    Originally Posted by theseeker2 View Post
    Originally Posted by Sharc View Post
    Nothing is below the legal limit.

    Image
    [Attachment 85071 - Click to enlarge]


    The capture is ok and shoudn't turn night/deep black into day/gray.
    You may increase the gamma somewhat in order to stretch the darks a bit, but there is nothing basically wrong with the capture levels IMO.
    If you prefer to tweak it do it in post.
    Do you think increasing brightness more will lead to more day/gray? I guess I can give it a shot and report back
    See jagabos post. Upload a video sample of your YUV capture rather than screenshots for better advice. My bad.
    Ok, see attached for 4 different video capture samples from roughly the same spot in the footage, from lower brightness to higher. The "undershoot" or whatever you want to call is definitely most pronounced in the last one, although I'm pretty sure the footage is still within the legal ranges in terms of blacks and whites.

    Capture 1: 01_cap.avi
    Capture 2: 02_cap.avi
    Capture 3: 03_cap.avi
    Capture 4: 04_cap.avi

    Any and all feedback is very welcome! Thank you!
    Quote Quote  
  14. Unfortunately, your capture device/drivers are crushing blacks below Y=16. That is, everything below 16 is being forced to 16. You're better off capturing with higher black levels and adjusting the levels later. You'll be able get more detail in the dark areas.
    Quote Quote  
  15. Originally Posted by jagabo View Post
    Unfortunately, your capture device/drivers are crushing blacks below Y=16. That is, everything below 16 is being forced to 16. You're better off capturing with higher black levels and adjusting the levels later. You'll be able get more detail in the dark areas.
    Good to know, thanks. I'm using the ATI USB 600. I have a Diamond VC500 on order, I will give that a try since it supposedly captures the full range 0-16.
    Quote Quote  
  16. The ugly dark shadows are actually caused by the "undershoots" (Y<16).
    Here an attempt of a fix based on your 04_clip.avi by "re-cropping" the shadows and recover few more details. Not perfect though.
    Code:
    V4=AVISource("04_cap.avi").converttoYV16(interlaced=true)
    v4corr=v4.levels(22,1.9,235,0,235,coring=true,dither=true)
    v4corr=v4corr.QTGMC(preset="fast").temporaldegrain2(degrainTR=2)
    sh=stackhorizontal(v4.subtitle("04_cap"),v4corr.subtitle("corr"))
    turnright(sh).histogram().turnleft()  #waveform monitor
    Image Attached Files
    Quote Quote  
  17. Originally Posted by Sharc View Post
    The ugly dark shadows are actually caused by the "undershoots" (Y<16).
    Here an attempt of a fix based on your 04_clip.avi by "re-cropping" the shadows and recover few more details. Not perfect though.
    Code:
    V4=AVISource("04_cap.avi").converttoYV16(interlaced=true)
    v4corr=v4.levels(22,1.9,235,0,235,coring=true,dither=true)
    v4corr=v4corr.QTGMC(preset="fast").temporaldegrain2(degrainTR=2)
    sh=stackhorizontal(v4.subtitle("04_cap"),v4corr.subtitle("corr"))
    turnright(sh).histogram().turnleft()  #waveform monitor
    That's actually pretty incredible compared to what I started with! Is that the complete AviSynth script that you included? I've used AviSynth for deinterlacing with QTGMC, but haven't tried anything else yet. Will have to study up on this
    Quote Quote  
  18. Originally Posted by theseeker2 View Post
    Is that the complete AviSynth script that you included?
    Yes.... any doubts?
    The essential is line number 2. The intention was to increase gamma to recover more details and structure from the darks, dithering to avoid banding (gaps in the histogram) and levels shift down again to force the original Y<16 (responsible for the ugly shadows at abrupt contrast change at edges) into out-of-gamut RGB (clipping). The rest is cleaning up by deinterlacing and denoising, and stacking SBS for comparison.
    Some rationale, some tinkering.... Can probably be fine tuned/improved.

    I don't know where the abrupt heavy undershoots at contrast edges (Y->0) came from. Baked into the source, capture device .....

    Edit: I just noticed that the SBS comparison is not based on the same framerate / number of frames due to the double rate deinterlacing. It affects the direct comparison view only, not the principle. A fix is on the way ....

    Edit2: So here again, synched (still I am not sure if the source is truly interlaced or telecined; I assumed interlaced as it is on old beta):
    Code:
    V4=AVISource("04_cap.avi")
    v4=v4.assumeTFF().converttoYV16(interlaced=true)
    v4corr=v4.levels(22,1.9,235,0,235,coring=true,dither=true)
    v4corr=v4corr.QTGMC(preset="fast").temporaldegrain2(degrainTR=2)
    sh=stackhorizontal(v4.changefps(59.94).subtitle("04_cap"),v4corr.subtitle("corr"))
    turnright(sh).histogram().turnleft()  #waveform monitor on top
    Image Attached Files
    Last edited by Sharc; 26th Jan 2025 at 08:19.
    Quote Quote  
  19. Here I used an Overlay() with an alpha mask to overlay a brighter version of video only where the black "shadows" were:

    Code:
    v4 = LWLibavVideoSource("04_cap.avi") 
    
    darkmask = v4.Invert().ColorYUV(off_y=-228).ColorYUV(gain_y=10000).Blur(1.5) # a mask of the darkest areas
    lighter = v4.ColorYUV(off_y=20).AddGrain(10) # a lighter version of the video
    Overlay(v4, lighter, mask=darkmask) # overlay lighter version only in the darkest areas
    
    StackHorizontal(v4, last, darkmask)
    TurnRight().Histogram().TurnLeft()
    I left the video otherwise intact so you could see that only the dark areas were lightened. From there you can adjust the levels, reduce noise, etc.
    Image Attached Files
    Quote Quote  
  20. Thanks jagabo, that is pretty nifty! I am grateful for you all sharing your expertise with me.
    Quote Quote  
  21. It might be possible to restore the actual picture information in those dark areas by using an even higher black level when capturing, and slightly modified script.
    Quote Quote  



Similar Threads

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