Go Back    Forum > Digital Video > Video Project Help > Capture, Record, Transfer

Reply
 
LinkBack Thread Tools
  #1  
08-16-2018, 07:53 PM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
I was doing a test capture when a dramatic scene change occurred. I had previously set levels to properly contain the histogram values within the safe dark and light boundaries for the prior scenes. Then the scene appeared that blew out the upper limit. I went back and studied the first part of the change, and while I was looking at the captured frame the histogram started changing. After about 20 seconds the values in the histogram had shifted to the left to bring the values within range. I assume this was the AGC on the VC500 card--true?

I captured two pictures of a single frame (field?) at the beginning of the scene change. They show the histogram change over about 20 seconds. Is there a way to change the AGC's affect on the video? (The image softness comes from the SnipIt tool.) How does this compare to the AGC on other capture devices?

8-16-18 vc500 capture before gain change.jpg

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.


Any idea what causes the saw tooth top edge on the histogram? The VC500 looks like my crosscut saw. the 2250 histogram looks like my rip saw.

Thanks,
Steve


Reply With Quote
Someday, 12:01 PM
admin's Avatar
Ads / Sponsors
 
Join Date: ∞
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
  #2  
08-16-2018, 08:37 PM
lordsmurf's Avatar
lordsmurf lordsmurf is online now
Site Staff | Video
 
Join Date: Dec 2002
Posts: 13,508
Thanked 2,449 Times in 2,081 Posts
Hard to tell without a clip. Maybe?

AGC is sudden brightness "correction" by a capture card. It makes the image worse, corrects nothing.

AGC manipulation was also one way how copy protection worked, and the AGC mess in capture cards is usually tied to this stupid trickery.

- Did my advice help you? Then become a Premium Member and support this site.
- For sale in the marketplace: TBCs, workflows, capture cards, VCRs
Reply With Quote
  #3  
08-17-2018, 02:18 AM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
Ok, I finally was able to capture a clip that produced similar brightness levels to the static images above. Don't laugh at my video choice. This was from one of my daughter's old tapes that I was using to practice on.

It took over an hour to get this clip. Every time I backed up the VCR to get the clip, VD or the capture card decided to set its own levels independently of the levels I set or the levels it picked previously. Sometimes the histogram indicated the indoor scene was overly bright where it was fine before. Other times the beginning of the bright outdoor scene did not send the histogram into brightness overload. During all of this I kept my adjusted levels the same for each capture. (On another note, VD had trouble figuring out what it was doing. Several times it presented an error message that another program was using the file VD was to write to. VD was the only program using the file--another frustrating idiosyncrasy.)

One thing I noticed when I played a longer section of the indoor scene where the histogram showed the scene as too bright, the levels slowly and automatically shifted left on the histogram to tone down the upper end of the histogram. The levels shifted enough to put the bright outdoor scene into range when it showed up. This was the same action I saw when I stopped on a frame that was too bright. The levels automatically shifted to bring the image into compliance.

Bottom line, two questions:

- what causes the levels to be different each time a video section is recaptured? This change each time makes it hard to manually preset the levels so nothing will go into overload during capture.

- what causes the automatic level changes during a scene--VC500 AGC or software?

Thanks,
Steve

Also, what causes the histogram saw tooth pattern?


Attached Files
File Type: avi 8-17-18 VC500 capture test 1.avi (72.02 MB, 10 downloads)
Reply With Quote
  #4  
08-17-2018, 02:54 AM
lordsmurf's Avatar
lordsmurf lordsmurf is online now
Site Staff | Video
 
Join Date: Dec 2002
Posts: 13,508
Thanked 2,449 Times in 2,081 Posts
In that sample, I don't really see AGC change.
Are you saying the scene brightness change from capture to capture? Can I get clips comparing that?

- Did my advice help you? Then become a Premium Member and support this site.
- For sale in the marketplace: TBCs, workflows, capture cards, VCRs
Reply With Quote
  #5  
08-17-2018, 12:38 PM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
The level in the clip would change over several seconds if I paused the video at the beginning of the bright outdoor scene. The histogram would slowly shift to the left while the video was paused. My first post shows images of the histogram when I first paused and about about 20 seconds later.

This action also shows up while the video is running if the level is consistently above the upper threshold. The histogram slowly shifts left until the upper level is well within the safe range.

I'll try to get clips showing the different levels from one recording to another. It was tricky getting the prior clip at just the right time. Also, VD has changed its mind today and is not showing the video while the histogram is on. I won't be able to accurately tell the levels from one capture to another until I can see both items.

Thanks for looking at this,
Steve
Reply With Quote
  #6  
08-17-2018, 01:56 PM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
EDIT: OK, STOP THE PRESSES. What was up is now down. What was left is now right. VD can show video and histogram at the same time! I didn't make any changes, it just decided to cooperate. To add more insanity to this work, the histogram no longer shifts on its own when encountering a long over exposed scene. Also, it does not shift when I pause on an over exposed image. The only thing that has happened since last night and today is the computer was shut down and I went to sleep. The VC500 was powered on all night even when the computer was shut down.

In addition, the levels I set yesterday to tame the over exposure are not sufficient today. Every time a bright object like a window in the background shows up it causes the histogram level to go out of bounds. Early yesterday it didn't. I feel like I am shifting between multiple universes.

I attached a clip from today that has the brightness shifting into overload every time a window shows up in the background. I cropped the clip for this test to get rid of the black and white edges (I know I need to take the cropping out when capturing my videos for real). I don't know that this is worth anything since the levels were not adjusted automatically.

My only course of action at this time seems to be set levels best I can and let-er-rip.

Thanks for looking at this,
Steve


Attached Files
File Type: avi 8-17-18 VC500 capture test 2 too bright.avi (96.26 MB, 3 downloads)
Reply With Quote
  #7  
08-17-2018, 10:33 PM
sanlyn sanlyn is offline
Premium Member
 
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,307 Times in 982 Posts
Quote:
Originally Posted by stevevid View Post
while I was looking at the captured frame the histogram started changing. After about 20 seconds the values in the histogram had shifted to the left to bring the values within range. I assume this was the AGC on the VC500 card--true?
False. The VC500 is not capable of AGC activity. Neither is Virtualdubb capture nor VCR playback. If there are AGC effects, they're in the original video. They are always a common annoyance with amateur cameras.

The sample clip isn't an example of AGC. It's an example of the way camera flare washes out contrast in an image. Later in the clip the flare goes away and the contrast range returns to normal.

It is, however, an example of other things. First, the flare is bright to the point of clipping beyond the safe zone of y=235. Second, the sample clip is an example of invalid cropping and nonstandard frame size. Hopefully that's not the frame size you captured. You apparently cropped black borders from the sides and bottom during capture. Your capture frame should be 720x480, including the original borders. Borders are always cropped and adjusted later during post-processing. Invalid cropping corrupts chroma response. Always use even numbers for YUV colorspaces, and always use even numbers for interlaced and/or telecined video. This video clip is hard-coded telecine, which for all practical purposes means it plays as interlaced in your VCR and TV and should be captured that way. Telecined video should not be deinterlaced. Instead, if you want progressive video from telecined sources you would use an inverse telecine filter (TIVTC) in Avisynth which removes duplicate telecined fields and restores the original progressive film rate of 23.976 fps. If you later want 29.97 fps playback restored, you can encode the video with telecine flags (aka 3.2 pulldown or some other form of pulldown depending on the original source frame rate).

Another downside of uneven cropping and nonstandard frame sizes is that encoders and authoring programs will choke on such frames. Yet again, many filters can't be used properly on frame sizes that are not mod-8 in all dimensions (meaning that the height and width must be evenly divisible by 8). Video dimensions that are odd numbers or not mod-4 can't be converted to YV12 for MPEG or h.264 processing. Also, the exterior walls of the curch's external stonework lines are screwed up and twittery on the downward camera pan, caused by cropping odd-numbered pixels from the height of yuv and interlaced/telecined video -- this can't be repaired after capture.

Quote:
Originally Posted by stevevid View Post
Is there a way to change the AGC's affect on the video?
No. Sometimes an auto filter such as hdrAGC or AutoAdjust can calm luminance pumping from consumer camera AGC "features" and autowhite "extras" (both are the work of the devil), but not by much. Auto features almost always do exactly what you don't want.

If you browse our capture and restoration forums you'll find AGC nightmares posted all over the place. Here's one example post that I can thank my sister for: Information Overload #5

And here's a tutorial post that has lots of jaggy-peaked histograms: Information Overload #2

And an example of a typical capture histogram that looks like yours:http://www.digitalfaq.com/forum/vide...html#post42215

This is normal for histograms unless they come from video having no differences between color changes, usually soft-focus or low-contrast scenes. The "dips" in the graph indicate lower amounts of certain colors or luma values, higher peaks indicate more instances of certain colors over others. Sometimes dithering filters can smooth things, sometimes not.
Reply With Quote
The following users thank sanlyn for this useful post: stevevid (08-18-2018)
  #8  
08-18-2018, 06:37 PM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
Thanks!

Quote:
Originally Posted by sanlyn View Post
False. The VC500 is not capable of AGC activity. Neither is Virtualdubb capture nor VCR playback. If there are AGC effects, they're in the original video. They are always a common annoyance with amateur cameras.

The sample clip isn't an example of AGC. It's an example of the way camera flare washes out contrast in an image. Later in the clip the flare goes away and the contrast range returns to normal.

It is, however, an example of other things. First, the flare is bright to the point of clipping beyond the safe zone of y=235.
The AGCish effects I was seeing were after/during digitization. I'm glad the capture device is not adding automatic "improvement."

Post #1 shows two effects I encountered and why I wondered if some form of AGC was involved during or after capture. The first image shows a change in levels I did not make. Before capturing this image, I cropped the video to get rid of the side bars and trash at the top and bottom. I then ran the video and set the brightness & contrast levels so all of the video contents I wanted were within the safe ranges including the change from indoor scenes to the bright outdoor scene. I planned to take the cropping out before my official capture so the format was correct. The item I found odd was that the video I captured after I set the levels showed out of bounds information I had taken out previously. My preset levels were still in the positions I set them in. Something else decided to shift the histogram values to the right.

In post #1, the two pictures are screen captures of one frame (field?) right after a change from an indoor scene to the bright outdoor scene. The first image shows the histogram right after the bright image showed up. The second image shows how the histogram changed over 20 seconds while viewing the same frame. The video was paused and not running. Something was changing levels and it was not me. It's as if some software AGC was at work. Unfortunately or fortunately I have not been able to reproduce this effect over the last couple of days.

Quote:
Originally Posted by sanlyn View Post
Second, the sample clip is an example of invalid cropping and nonstandard frame size. Hopefully that's not the frame size you captured. You apparently cropped black borders from the sides and bottom during capture. Your capture frame should be 720x480, including the original borders. Borders are always cropped and adjusted later during post-processing. Invalid cropping corrupts chroma response. Always use even numbers for YUV colorspaces, and always use even numbers for interlaced and/or telecined video. This video clip is hard-coded telecine, which for all practical purposes means it plays as interlaced in your VCR and TV and should be captured that way. Telecined video should not be deinterlaced. Instead, if you want progressive video from telecined sources you would use an inverse telecine filter (TIVTC) in Avisynth which removes duplicate telecined fields and restores the original progressive film rate of 23.976 fps. If you later want 29.97 fps playback restored, you can encode the video with telecine flags (aka 3.2 pulldown or some other form of pulldown depending on the original source frame rate).
Yes, you're right. The clip includes cropping. I already planned not to use cropping when I record a whole video. This video is just one I am using to learn with. The videos I will be capturing for real are family, home videos of varying quality and I didn't want to stress them until I am ready for my official captures.

I posted the clip with cropping to just focus on the video content without the additional trash. Should I not do that and always post clips without cropping?

Quote:
Originally Posted by sanlyn View Post
Another downside of uneven cropping and nonstandard frame sizes is that encoders and authoring programs will choke on such frames. Yet again, many filters can't be used properly on frame sizes that are not mod-8 in all dimensions (meaning that the height and width must be evenly divisible by 8). Video dimensions that are odd numbers or not mod-4 can't be converted to YV12 for MPEG or h.264 processing. Also, the exterior walls of the curch's external stonework lines are screwed up and twittery on the downward camera pan, caused by cropping odd-numbered pixels from the height of yuv and interlaced/telecined video -- this can't be repaired after capture.
Thanks! I had not read about mod-8 (mod-4?) and problems with odd numbers.

Quote:
Originally Posted by sanlyn View Post
Sometimes an auto filter such as hdrAGC or AutoAdjust can calm luminance pumping from consumer camera AGC "features" and autowhite "extras" (both are the work of the devil), but not by much. Auto features almost always do exactly what you don't want.

If you browse our capture and restoration forums you'll find AGC nightmares posted all over the place. Here's one example post that I can thank my sister for: Information Overload #5

And here's a tutorial post that has lots of jaggy-peaked histograms: Information Overload #2

And an example of a typical capture histogram that looks like yours:http://www.digitalfaq.com/forum/vide...html#post42215

This is normal for histograms unless they come from video having no differences between color changes, usually soft-focus or low-contrast scenes. The "dips" in the graph indicate lower amounts of certain colors or luma values, higher peaks indicate more instances of certain colors over others. Sometimes dithering filters can smooth things, sometimes not.
Again, thanks for the information and links. I hadn't run into auto fixers. I'll view them with suspicion until I'm more knowledgeable.

The other links are great. I haven't fully absorbed them yet, but I really appreciate the work you put into the responses.

Steve
Reply With Quote
  #9  
08-18-2018, 10:38 PM
sanlyn sanlyn is offline
Premium Member
 
Join Date: Aug 2009
Location: N. Carolina and NY, USA
Posts: 3,648
Thanked 1,307 Times in 982 Posts
Quote:
Originally Posted by stevevid View Post
The AGCish effects I was seeing were after/during digitization. I'm glad the capture device is not adding automatic "improvement."

Post #1 shows two effects I encountered and why I wondered if some form of AGC was involved during or after capture. The first image shows a change in levels I did not make. Before capturing this image, I cropped the video to get rid of the side bars and trash at the top and bottom. I then ran the video and set the brightness & contrast levels so all of the video contents I wanted were within the safe ranges including the change from indoor scenes to the bright outdoor scene. I planned to take the cropping out before my official capture so the format was correct. The item I found odd was that the video I captured after I set the levels showed out of bounds information I had taken out previously. My preset levels were still in the positions I set them in. Something else decided to shift the histogram values to the right.
Borders and head-switching noise are elements that affect the histogram, not the image. Most people do as you plan: crop temporarily to measure the image content without the garbage, then restore the borders so that the frame can be configured properly after capture.

Quote:
Originally Posted by stevevid View Post
In post #1, the two pictures are screen captures of one frame (field?) right after a change from an indoor scene to the bright outdoor scene.
They are images of two frames, each containing two fields.

Quote:
Originally Posted by stevevid View Post
The first image shows the histogram right after the bright image showed up. The second image shows how the histogram changed over 20 seconds while viewing the same frame. The video was paused and not running. Something was changing levels and it was not me. It's as if some software AGC was at work. Unfortunately or fortunately I have not been able to reproduce this effect over the last couple of days.
I can't tell you what happened there. Too bad the posts omitted elements that might have explained the event.

Quote:
Originally Posted by stevevid View Post
I posted the clip with cropping to just focus on the video content without the additional trash. Should I not do that and always post clips without cropping?
Always post unaltered samples. I could have demonstrated how to configure borders and maintain the original image aspect ratio while working the borders but there was no way to determine the shape and extent of the original borders. Besides, the odd cropping had its own effects that couldn't be corrected for the demo.

Quote:
Originally Posted by stevevid View Post
Again, thanks for the information and links. I hadn't run into auto fixers. I'll view them with suspicion until I'm more knowledgeable.
There are several auto filters in Avisynth and a coup-le in VirtualDub.The two most workable auto levels are Avisynth's HDRagc and AutoAdjust. In any case their use requires patience and, ultimately, they're often discarded as not being very cooperative and just as annoying as AGC in cameras. Still, on rare occasions, they can help if you learn how to set them up.
Reply With Quote
The following users thank sanlyn for this useful post: stevevid (08-19-2018)
  #10  
09-15-2018, 05:38 AM
lordsmurf's Avatar
lordsmurf lordsmurf is online now
Site Staff | Video
 
Join Date: Dec 2002
Posts: 13,508
Thanked 2,449 Times in 2,081 Posts
I've seen some clips from stevevid in other posts, some comparing capture cards, and I may have to agree with him. It does look like some sort of very slight AGC exists, at least in his card.

As we've seen with other hardware, perhaps he has an earlier or later card than you do, sanlyn. Or different altogether, yet still carrying the VC500 model number.

I guess the only thing left is to get to post a picture of the VC500, make sure it really is one (and that he wasn't sold something else).

So, do that stevevid.

- Did my advice help you? Then become a Premium Member and support this site.
- For sale in the marketplace: TBCs, workflows, capture cards, VCRs
Reply With Quote
  #11  
09-15-2018, 02:27 PM
stevevid stevevid is offline
Free Member
 
Join Date: Jul 2018
Location: Dallas, TX
Posts: 124
Thanked 12 Times in 9 Posts
Here they are.

VC500 front.jpg

You must be logged in to view this content; either login or register for the forum. The attached screen shots, before/after images, photos and graphics are created/posted for the benefit of site members. And you are invited to join our digital media community.


I scrapped this VC500 because of two things:

- Annoying sometime active slooooow AGC
- Catastrophic arbitrary level changes. My settings remainded the same, but the VC500 would sometimes change what were safe levels into levels that slammed into both the 0 and 255 walls. Never found why or how to fix it.

I switched to my 2250 card and it worked flawlessly. Levels always stayed where I put them, no AGC activity, and all captures came through without dropped or inserted frames (even after 2 hours). A very solid device.

The only annoying part of the 2250 with VirtualDub is getting to see the video and histogram at the same time. I got it to work after I stood on my head while spinning and playing the 1812 Overture on a piano and harmonica.



Last edited by stevevid; 09-15-2018 at 02:37 PM.
Reply With Quote
The following users thank stevevid for this useful post: archivarious (01-20-2021)
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
Capture whole tape with one set levels, or different levels per scene? stevevid Capture, Record, Transfer 5 08-22-2018 11:29 PM
Setting up ATI 600 USB levels? Zerowalker Capture, Record, Transfer 2 07-06-2018 10:32 AM
Changing Title Tags after site is indexed by Google? Mr. Rey Web Development, Design 4 10-05-2013 04:37 AM
Won't load new website, keeps loading old one after changing ns1/ns2 ? cyber-junkie Website and Server Troubleshooting 7 07-19-2012 01:59 AM
DVDWS2 Time of Mpg issue after changing to VideoTS Konfusion Author, Make Menus, Slideshows, Burn 10 12-14-2007 10:25 PM

Thread Tools



 
All times are GMT -5. The time now is 12:46 AM