Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

heavy vertical line artifacts on EPIC shots

S

Simon Sturzenegger

Member
Joined
Aug 23, 2009
Messages
20
Reaction score
0
Points
0
Location
Berlin
Hi there,

On some shots we did today we discovered heavy vertical line artifacts throughout the whole picture. It's not visible on all the clips on the card so I guess it has to be a problem with the internal camera debayering? Any clues about this problem?
We are shooting on the RED Epic firmware build 2.0.5

Here is a screenshot of the problem. You can see the effect very good on his nose and also the horizontal lines throughout the whole pictures...

http://dl.dropbox.com/u/679472/epic_lines.tiff

Any help would be appreciated!


Simon
 
Can you post your pic again, it seem like it's a deak link. I met the same problem before, so I need to see your picture.
 
We had an issue with vertical lines in Epic footage a while back. It was evident in export from RCX but not in the preview. It exported from Resolve without a hitch. The solution was to update to a newer version of RCXpro.

If that doesn't help: How are you viewing the footage (RCX, Resolve, etc), are you using a REDrocket, and have you tried transcoding it?
 
I had a vertical black line on Red One M-X before. I can't see any vertical black lines when I play back in camera, I can see it in quicktime or RedCine-X. When I transcode R3D to QT Apple Prores422, it's still there.
 
new dropbox link

new dropbox link

Sorry changed the dropbox link!

Here you find a ZIP package with three screenshots in it: http://dl.dropbox.com/u/679472/epic_lines.zip

I am viewing the footage in RCX-Pro with a RedRocket on a MacPro. I did not try to transcode it yet, but I will the next day I'm on my workstation, thanks for the tip! Would be strange though if it would be "normal" on exported ProRes, because the lines only show up on five clips and not on all of the footage...
 
I just had this exact same problem on a greenscreen shoot. One mag exhibited this issue onall the clips recorded on it.

The other 20 or so magazines were 100% fine. Maybe a bad format? Annoying because I missed this on set, as I scrubbed through at 1/4 res to check, and it looked just like any other clip at 1/4. Exported OpenEXR for the Nuke guy on set and he didnt notice either.

I sent the clips to RED for an answer, still waiting to hear back.
 
Hi Simon,

I looked at your samples. That looks like what we experienced a while back. The vertical steps were visible on hard edges but also seen as a light pattern in highlights. In our case it happened on a single clip in the midst of a full day of shooting. All the other clips were fine. Red support recommended updating the firmware, which we did not do at the time because a release version was not available - and we haven't seen that particular issue since. This was a few firmware versions back. These types of issues are frustrating because they appear to happen randomly and quite infrequently, so it's best to send Red a log file from the camera and a few seconds from an R3D so they can take a look at it.
 
Thats kind of what it looks like when viewing footage in 1/8th or 1/16th in RedcineX. Thats weird...
It's almost like a bad debayer.
Click to expand...

Yeah you're right.. If I look at it in 5K Full Debayer 100% it looks really bad. Not usable IMHO.

Thats kind of what it looks like when viewing footage in 1/8th or 1/16th in RedcineX. Thats weird...
It's almost like a bad debayer.
Click to expand...

Just transcoded it to ProRes4444 and it looks the same. It's not that visible on 1080p but still... Exported DPX now and trying to get a 2K screening in the post facility

Thanks for all your feedback this far!
I just uploaded a R3D screenshot for you to take a look at it in full res: http://dl.dropbox.com/u/679472/A058_C008_0307N9.0001653F.R3D
 
it looks like the certain frames were corrupted in camera. I think the debayer stuttered or something. Weird.

It also looks like a bad black shade. Theres a couple blue dead pixels (possibly noise, but theres only like 6 I could find, and they are only visible in the darks.) in the dark part of the wall behind him when you zoom closely.
 
WHATTTTT IS THIS!!! just got a call..

my camera is on the field.. they saw the vertical lines

not good... not good.. need an answer
 

Attachments

  • A010_C007_0223CB_001_1324381_bad.jpg
    A010_C007_0223CB_001_1324381_bad.jpg
    66.4 KB · Views: 0
Last edited:
Contact RED directly for this one, guys. It's generally from a bad cable short or an electric discharge onto the camera body from an external source.
 
Brook Willard said:
Contact RED directly for this one, guys. It's generally from a bad cable short or an electric discharge onto the camera body from an external source.
Click to expand...
Do you mean like not grounding yourself before touching the camera?
 
Has anyone found a solution to this problem or was this posted in another thread? Any follow up information would be great as I'm experiencing similar problems.
 
Kieran Moreira said:
Has anyone found a solution to this problem or was this posted in another thread? Any follow up information would be great as I'm experiencing similar problems.
Click to expand...

Hey Kieran,

If you can go ahead and send me an email with a LOG file, and get me an R3D, we can take a look for you. Nate@RED
 
Nate Heartt said:
Hey Kieran,

If you can go ahead and send me an email with a LOG file, and get me an R3D, we can take a look for you. Nate@RED
Click to expand...

Hi Nate! A client gave us footage with this problem. We're in the middle of a very tight post timetable and it looks like they don't have time or resources to do a reshoot. Is there a solution to cure the footage or are they essentialy f*ck*d and stuck with this problem? The camera supplier says that they are running the latest firmware on the EPIC.
 
Hi Jam,
I had this happen on a shoot almost a year ago and came up with a fix in After Effects that significantly suppresses the problem. It's not perfect, but at 1080p downscale, the problem pretty much disappeared in the footage we repaired. Send me a PM with your e-mail address and I'll send the AE CS6 project file. It operates on the original 5K footage and reorders 1-pixel wide columns into a correct order to eliminate the striations. One important note is that the render must use 'current' quality in render settings. Setting to 'best' quality will cause anti-aliasing in the pixel grid and make the whole thing not work.

Here are 5k blowups at 200% before and after:

Before
before.jpg


After
after.jpg


Thanks,
Evan


Jam Jimenez said:
Hi Nate! A client gave us footage with this problem. We're in the middle of a very tight post timetable and it looks like they don't have time or resources to do a reshoot. Is there a solution to cure the footage or are they essentialy f*ck*d and stuck with this problem? The camera supplier says that they are running the latest firmware on the EPIC.
Click to expand...
 
I wrote a very convoluted and speculative post and lost it. Here is the essential guts of my way out theory on cause if none other can be nailed down and no solution found.

Maybe examine the possibility of a vibrational couple of acoustic or direct mechanical origin which has a harmonic relationship with the frame rate. Something like the tripod is standing on a floor being shook from a nearby chiller or magnetic ballast. Fan motor shaking a lens which has a loose element or iris assembly inside, cheap mattebox and followfocus on thin rods getting into a harmonic vibration in a torsional motion - sideways movement transmitted back to the opitcal axis causing a vertical artifact. Monitor vibrating on its mount from constant acoustic or mechanical highfrequency disturbance. The camera internal fan might just be capable of generating a mechanical vibration from an out of balance state or acoustic pulses from the fan disk itself.

The magnitude of any such movement would be so small but at a pixel row pitch magnitude, five microns or so, it might be enough introduce an artifact if the lens optical axis is being disturbed by vibration ampified by a mattebox on skinny rods.

It would occur so incredibly rarely and randomly as to be near impossible to replicate in testing unless all the cagework, accessories, fan speeds at harmonics of frame rate were repeated.


Where does this random thought come from. I have seen something like this previously when using a home-made groundglass adaptor with a shaky disk on interlace cameras. Odd and even interlace lines would differ in brightness and colour density. De-interlacing seemed to fix this but introduced a twitter or weaving artifact on horizontal edges. A vibrational artifact should be most apparent at longer focal lengths. It may also be restricted to wide views if the movement is so fine as to coincide with just two pixel rows or in the case of my PD150/AGUS35, confined to just two interlace lines.

http://www.youtube.com/watch?v=AQGsJStUHAE

In the first wide shot of this clip, you will observe the vertical weave artifact on a horizontal edge which remained after I cleaned the footage up as best I could. There is no need to view the rest of the clip. The vibration was random. It did not occur in my closeups with longer lenses. There was some camera shake as the balcony floor was not firm. Yours of course affects vertical lines.

My post-process path was the old Premiere 6.1, de-interlace, colour and de-interlace in After Effects 5. Subsequent cleanup was attempted in Premiere CS5. A one-frame offset 3 track layer stack was introduced to minimise sensor gain noise. That is where the smear trails come from. Coding for YouTube caused the bad banding artifact.

I don't think deinterlacing will help you as your artifact if of mechanical origin appears to be induced by horizontal movement and your footage is already progressive anyway. Rather than using downsampling technicques which affect all pixel rows and sacrifice sharpness, my attempt at a fix for the finer artifact might be to attempt to layer three identical clips, offset two, by respectively one and two horizontal pixel rows or more if the span of the artifact is greater. I would then either try frame blend or if that does not work, slide each of the pixel offset clips one frame later than the lowermost clip, give the top two 33% transparency and render at original pixel count, ie., if it is 5K, render at 5K. There will be motion trails which mean dynamic camera moves will suck to perdition. This fix will only work on shots with little movement. Offset layer stacking is handy for reducing gain noise but there are likely plug-ins which do the job much better these days. It might be a trick worth trying to recover some detail in macroblocked blacks, if the macroblocking is suffiently variable over short periods.


What is good for MiniDV/DVCAM is likely another matter with 4K+ footage. It is likely I am sending you on a goosechase but hopfully it is a line of investigation you can eliminate.
 
Last edited:
You must log in or register to reply here.
Back
Top