Black Shading Noise Real-time Indication

[Jake Bastian]

How does:

My recommendation is to blackshade:
1) upgrading to new firmware
2) big ambient temperature changes
3) long exposure times (longer than ~1/15th sec)
4) very short exposure times (shorter than around ~1/1000th)
5) beginning of a project
6) as often as you feel like

Remember to black shade at default after long/short exposure black shade when your finished.

More convenient/flexible black shading coming later.

Not answer your question? He's telling you exactly when to black shade so you don't waste your time.

 

[Deanan]

Good Afternoon Deanan,

I'm in New Zealand for 3 weeks of shooting. I really need to know what is the total RED technical / testing knowledge on Black Shading as I have outlined above. I'm not sure your reply: http://www.reduser.net/forum/showthr...l=1#post867800 is a response to this request since it does not cover my questions and is not comprehensive in topic coverage.

If RED has not conducted any organized testing on Epic / Black Shading involving changing frame rates / RC / / Resolution / Shutter then please let me know so I know how to spend my time tomorrow (location shooting vs testing in the hotel room). If there has been structured testing at RED then please forward that information so I can make a productive day of tomorrow and return with quality, noise free footage.

Please let me know either way.

Respectfully,

-Josh

Blackshading is the same as it was for epic and red one so that's about 5 years worth of testing both internally and field experience.

Sensor temperature and exposure time the two factors that influence blackshading under normal use.

 

[Brook Willard]

Deanan knows stuff. Just saying.

 

[Zeb B]

Good Evening Deanan,

So I understand, from what has been stated above, that switching Frame Rates and Resolution up and down will have no effect on the noise recorded. From the little observation (from short-term Epic ownership) I have had on the topic I suspect that moving between Frame Rates and Resolution in fact induces un-monitored noise (only shows up in CineX after acquisition). I will spend some time tomorrow to research this hunch and try to rule out Shutter Speed changes since noise is critical to quality and can make my acquisition footage worthless. My work calls for quick and constant switching between Frame Rates and, sometimes, resulting Resolutions - which is exactly why the Epic was the best fit for me. If there is a requirement for a 10+ minute recalibrate between switching of the FR/Rez parameters then that is a different value proposition.

Am I to also understand that my request for further information other than your above 6 point post will not be forthcoming? If so, why? I mean if there is 5 years of science on the topic why not share beyond the very brief general info post? A technical issue that directly drives image quality would seem to be something to share with those like me who invested heavily in Epic for the specific purpose of producing the highest quality images. I would ask respectfully whoever has this very important information please format it into a White Paper style document and release it to Epic owners so we can benefit from the wisdom and experience and add to it as experience provides more understanding of the topic. This seems a very reasonable request.

On your 6 points - 1-3 and 6 are basic and easily understood. Questions on 4 & 5:

4) very short exposure times (shorter than around ~1/1000th)
Does that mean only when I move from a slower shutter speed to 1/1000th I need to Black Shade? How about back down from high to low? How about moving from 1/1000 to 1/2000?

5) beginning of a project
I'm not sure what "project" means in this context. Does that mean Black Shade per power-up, per scene, per day of shooting or per the entire project to include any camera settings specific to the project?

Thank you for you time,

-Josh

 

[Zeb B]

Tested for a couple hours this morning. If the vertical gray bar left of the histogram is noise - I have plenty of it. These clips are shot just after Black Shade Calibrating at Current FPS / Shutter. Doing a second test using Factory Default B/S. I would love to have some TEAM RED input as to what might be going wrong here.

 

[Zeb B]

Next test:
Black Shade Calibration at Factory Defaults. Still have unexplained (and non-displayed on 5" Touch Screen level bars) noise at lower frame rate settings. Very odd that as the frame rate and compression increases, the gray noise bar decreases. If there a way to monitor this gray bar noise before/during acquisition?? How do I eliminate this noise completely? This noise is a critical quality issue for me.

 

[Zeb B]

Next test: BS at Factory Defaults at with camera set at 5K then shooting at 4K. Noise spikes up. This seems to suggest that moving from one resolution to another creates un-monitored noise.

 

[Zeb B]

Now the most puzzling of the tests. Black Shaded at Factory Defaults with camera set to 5K, shot the first 5K test clips above (post #26) then shot the 4K test clips in post #27. Then shot again at 5K - noise spiked up substantially suggesting that moving between resolutions requires a new Black Shade Calibration process. I would like to have this test series varified or technically explained / rebutted to help me understand what is happening. Using 1/2 of a REDVOLT charge (10+min) to recalibrate between each resolution change is not always practical in the field.

 

[Tyson Perkins]

+1

Would really like to see a more thorough response from Red. I don't know how many times I've read of this issue to very little reply from Red.... I am no trying to insinuate that Red is hiding something or trying to avoid this particular topic for any given reason, but just that it seems that this particular topic gets a minute amount of official acknowledgement when compared to the majority of others.

 

[Zeb B]

Ran the B/S Calibration twice. Both times it took around 24 minutes. That excludes the use of REDVOLTs to power the brain during calibration since that is at or beyond the maximum run time for them. This complicates field use of a lightweight Epic.

After Calibration an option comes up in the Maintenance section entitled "user HS". This is different than the "user" title shown on page 125 of the latest Epic manual. Not sure what this means if anything.

If anyone has anything to say on this topic I would love to hear. Seems like potential Scarlet features and other pithy subjects gets hundreds of replies - one of technical / quality consequence gets very little. Where are the experts in REDland or is this primarily a social site to reaffirm each others support for the brand?

I need help please.

 

[Paul Leeming]

Josh, that grey bar has nothing to do with recorded noise, but rather the amount of processed image (with current settings) that dips into pure black. As long as your on camera exposure didn't show up any purple (using the exposure check function) then your RAW noise should be minimal to non-existent.

The grey bar is a quick grading tool to let you know you're pushing the blacks to 0,0,0 as opposed to how much noise/grain was recorded.

HTH

Cheers from Tokyo,

Paul

 

[Sam Winzar]

Paul is correct. There is no "noise indicator". The easiest way to see if there is noise, is just look at the image. None of those images look noisy (but it's hard to tell as they're small screen grabs of 1/8th quality). If you're worried about not noticing noise due to the down sampling of the the monitor image, use the 1:1 function and you can see pixel for pixel how noisy the image is. Just remember that as you down sample a lot of that will go away.

 

[Justin O'Neill]

The grey bar on the left in those shots Josh is posting just don't make sense to me. When I look at similar shots we've shot on the Epic I really have to crush the blacks in order to get the grey bar to look like that. Most of those images looks very flat to me...does that look right to everyone else?

 

[Zeb B]

Paul, Sam and Justin,
Thank you for your valuable input. I ran some real life landscape scenery tests this afternoon and it's freaky how the gray bar spikes on some real flat looking images. I am checking clipping on both high and low with the Purple/Red preview. From some of the files it is real tough to grade the gray bar down. Adding Saturation or other tonal adjustments pegs the bar to the max. Is this normal? Doesn't seem right to have crushed blacks right out of the camera in RAW - I would expect that from a lesser camera compressed file.
Thanks again,
-Josh

 

[Paul Leeming]

Maybe the grey bar is behaving buggy in build 7? Why not render out a TIFF and import that into Photoshop and double check the level distribution there, as a kind of sanity check.

At the end of the day, if the image looks good then I'd say there's nothing to worry about. If you can share an R3D snapshot we can independently verify for you as well.

Cheers from Tokyo,

Paul

 

[Zeb B]

How do I upload a R3D snapshot? Flickr no likie RED files.

Where is there a RED CineX Operation Manual that describes what the interface means and how to use it?

It looks to me as if my Epic Green Channel is jacked. Take the slider to the far right (either by Exposure, ISO or FLUT) and weird yellow dots remain - which seems to be the "crushed" areas. Try to add even a small bit of saturation to this very flat image and the crushed black gray bar indicator spikes up to maximum. I have worked with Photoshop RAW images from my Canon and PhaseOne and have not seen anything like this. Does this mean my Epic sensor has a defective green channel or is there a software issue. Had to tell since there is no documentation or self help for anything.

See the very underexposed fireplace shot at the bottom. No green component = no crushed blacks out of camera. Is it the massive amount of noise in that low light shot that prevents the blacks from being crushed or the lack of green? How to I field test what I suspect? Being in New Zealand with very slow $30/day Internet . . . any self-help or things to check would be helpful. My primary shooting starts tonight and goes for 3 weeks from there. Thank you in advance.


As shot - Very flat image, hump in the middle, no Red/Purple clip indication on 5" Touch Screen, no Red Underexposed level vertical bar warning = crushed blacks out of camera:

Max Exposure = weird yellow dots:

Max FLUT:

Add just a small bit of Saturation - "crushed blacks" indicator spikes to max:

Shot in low light - no green component = no crushed blacks even though the Purple underexposure warning was pegged:

 

[Tim Sutherland]

Josh, the thing about scope is, nobody is going to watch the scopes in the end, only the picture. If you are grading based on the grey bar, and it seems wrong, test some other stuff.

Did you try different builds of rcx? Did you take the image to PhotoShop as Paul suggested? If the image you're seeing doesn't seem to match at all with the scopes you are seeing, then it seems like either monitor calibration or scope performance might be off.

I also noticed that all your tests were shot between 10:1 and 13:1 compression, which seems high for 5k. Did you try anything else?

it looks to me like something is up, but it doesn't look like noise is your problem. like others said, the gray bar isn't a noise indicator, it shows how much of your data is in the shadows.

Tim

 

[Zeb B]

Hey Tim,

Thank you for the input. I have no idea what the Gray Bar represents: Critical to Image Quality or Useless Information. If someone has an official document from RED as to the purpose of this bar (and the entire CineX interface for that matter) I would greatly appreciate posting it. I shot a bunch of footage in Napa a week ago that showed the same inexplicable gray bar phenomenon. The well exposed, flat footage had a bunch of noise in the blacks, making some of the footage unusable, so I don't think this is just a scope issue. All indications point to a Black Shading, hardware or firmware issue.

I would love to hear from RED as to how, based on the above thread, I can field test my camera to see if there is something mechanically wrong so I don't waste time shooting defective footage. What would they do at inbound QC to quickly check functions?

On the compression ratio: does 10:1 or higher crush blacks?

Here is the same frame from CineX and Photoshop. Not sure if this is a test of anything since one is a Tiff and one a R3D preview. Processing out to Tiff introduces another variable in the test chain.

Thank you for the assistance. I hope to use the Epic starting tonight since I carried 100lbs of shit halfway across the planet to do so.

 

[Les Dittert]

The dark subtraction is basically taking some averaged black images from the sensor , and storing this image. Then, as the sensor captures your taking images, the dark image is subtracted pixel for pixel from the taken image. This partially corrects for FPN , or fixed pattern noise. But it really isn't noise so that is misleading. It's just a fixed static pattern, so it does not 'buzz' like grain or usual sensor noise. Some folk call it a 'screen door' pattern, because it can look like the image is filmed through a screen. As soon as the camera moves a little, you see a pattern that stays put on the image, almost like a dirty screen.

Anyhow, you should really only see this in dark stuff, say if you gamma it up a bunch to see into the blacks.

Red is probably also doing the gain of each pixel, stored in the cameras rom, along with defect pixel map.

I've developed film scanners a decade ago , same stuff happens there.
-Les

 

[Tim Sutherland]

If you start crushing the blacks, does the noise start to go away? What if you set ISO to 320 and grade from there?

Compression ratio doesn't affect the contrast, but those are highly detailed shots which will show the weakness in higher compression, most notably with noise in the shadows.

I would probably go with something more like 7:1 or 6:1 for highly detailed scenes like this. A bad or lack of black shading might be the cause, but I would probably suspect compression first.

Tim