Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Head clog glitches or something else? #194

Open
briancash opened this issue Dec 14, 2017 · 12 comments
Open

Head clog glitches or something else? #194

briancash opened this issue Dec 14, 2017 · 12 comments

Comments

@briancash
Copy link

Hi,

Myself and @kieranjol are looking to refine our glossary of dv errors before we begin ingest on a project. One error that keeps showing up when we test on QC Tools and DV Analyzer is an element of head clog. Given the example used on the AVAA site- are the small chequered flag glitches (highlighted in red) considered to be 'head clog' or they a separate issue that can occur outside of head clog? If so is there a suitable term for these?

image

Example we found:
image

Regards,

Brian

@ablwr
Copy link
Collaborator

ablwr commented Dec 14, 2017

You should submit that glossary to AVAA!

I tried to speculate and come up with an answer for you, but I can't confidently make an assertion here. I think it certainly could be a head clog or specific decay on the tape itself.

@bturkus
Copy link
Contributor

bturkus commented Dec 14, 2017

I also have no idea what to call this particular error, but thanks for submitting something unknown to the issue tracker. the system works! now we just need to get other people to use it.

@kieranjol
Copy link
Contributor

kieranjol commented Dec 14, 2017 via email

@briancash
Copy link
Author

I'll drum up a few more examples with the corresponding Dv analyzer info and update the thread tomorrow. Thanks for the reply @bturkus

@ablwr
Copy link
Collaborator

ablwr commented Dec 14, 2017

A closer crop might be helpful too because the resize causes compression artifacting?

@kieranjol
Copy link
Contributor

@ablwr would zooming in using QCTools and then taking a screenshot work?

@ablwr
Copy link
Collaborator

ablwr commented Dec 14, 2017

sure

@michaelangeletti
Copy link

Hi guys,

I don’t have definitive information regarding the artifact Brian inquired about, but in our production database at SMPL, I put those types of checkered flag pixel artifacts into the “DV dropout” column. I usually see them appear alongside the very common 32h x 8v pixel dropout configuration containing color patterns retrieved from other parts of the frame or previous frames. I think those checkerboard errors are usually much smaller than 32 x 8, and if they are black and white, maybe it indicates a reduction in the sampling of those groups of pixels from 4:1:1/co-sited 4:2:0 to 2:0:0 (just guessing here). Maybe it’s worth an email to DV guru Adam Wilt?

I think this highlights some of the challenges around using compressed still images to illustrate the subtler variety of artifacts (IIRC, we resized/resampled the original 720x480 AVAA example exports to square pixel, deinterlaced, moderately compressed jpegs… yuck). When I made that entry in the old MediaWiki AVAA, I think I was looking for interesting/dramatic examples of DV head clogs that would most effectively show the artifact, but sadly these other smaller artifacts within the frame went unaddressed. I will see if there are any other examples of this as I'm in the process of digging up the existing video examples for accessioning into the Stanford Digital Repository.

@briancash
Copy link
Author

briancash commented Dec 15, 2017

Thank you @bturkus , @ablwr and @michaelangeletti for your responses. I have attached zoomed in screenshots of the errors from Qc Tools including one glitch that is in colour (not sure if it's directly related to the other checkered flags). It would be handy to get some kind of terminology for these as they seem to pop up fairly regularly in this current project. If they fall under 'DV dropout' as per Michael's post above then at least that's something!

Many thanks,

Brian

screen shot 2017-12-15 at 12 41 17
Arb bits :7
Arb Bits Info: Not Sequential
Video Error Concealment: 54.00% (810 A errors)
screen shot 2017-12-15 at 12 50 25
Arb bits: 5
Arb Bits Info: Not Sequential
Video Error Concealment: n/a
screen shot 2017-12-15 at 12 52 17
Arb bits: 3
Arb Bits Info: Repeated
Video Error Concealment: n/a
screen shot 2017-12-15 at 12 53 32
Arb bits: 3
Arb Bits Info: Not Sequential
Video Error Concealment: n/a

@bturkus
Copy link
Contributor

bturkus commented Dec 15, 2017

Hi everyone,

If we end up in a place where we have to create our own terminology, I'm all for "DV Checkerboard Error." But, before we head down that path, let me reach out to Adam Wilt to see if he can shine a little light on this for us. It might not work out--the contact page on his site explicitly states, "Sorry, I don’t have time to answer general DV questions, and I flat-out refuse to tell you which camera to buy!"--but this is way more than a "general" DV question, right?

While on the subject of DV, I do have a few other questions/ideas:

(1) It looks like AVAA has two different DV tags--DV and DV25--and I'm wondering: can we collapse the categories, or is there a particular reason for keeping DV25 separate?
(2) We don't have examples for either of DV25 errors currently listed--Sub-sampling artifact and Mosquito Noise. Maybe we can all prioritize trying to find these? I've got large institution rights-related issues to work out before I can begin posting clips--and, of course, I don't really know what these artifacts look like--but there might be some hope...
(3) Has anyone given any thought to using FFmpeg to visualize DV errors in new ways? I believe @dericed has used the codecview filter to "unconceal" DV concealment errors. It's a very cool effect, though I can't find a clip on his YouTube page. I've played around with some of the commands found here (https://trac.ffmpeg.org/wiki/Debug/MacroblocksAndMotionVectors), but I mostly ended up with a bunch of confusing arrows zigging and zagging all over the place (like the 3rd pic of the bird).

Just a thought, but I know we can harness the powah of FFmpeg to teach people more about DV.

@dericed
Copy link
Member

dericed commented Dec 15, 2017

I'd prefer to reference discrete cosign transform (DCT) in the error name rather than the word checkerboard since it would point people in a more helpful direction. Also only 1/8 of those errors would appear in a checkerboard pattern.

@briancash
Copy link
Author

Having just googled to learn what DCT is... thats a good description for those errors

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

6 participants