Can someone explain this exposure situation for me?

[Stephen Pruitt]

Okay. . . there seems to be an inconsistency with Build 30's metering.

On the RGB graph below at the bottom, my exposure was definitely "exposed to the right," but NONE of the last goal post showed any red at all. Totally clear on that last segment.

The problem is that both the traffic light and the exposure bars were lit up fully. I thought that meant clipping. How can something have no clipping in one place and clipping in the other? And, yes, I was viewing RAW when this was observed. Personally, I would like for my meters to be consistent one to the next, as, in this case, it would make sense to me that they should be telling me the very same thing (except, of course, that they weren't!).

What is the point of the goal posts if they aren't showing us real noise or real clipping? If that is the accurate meter, then what is the point of the traffic light and the exposure bars if they aren't giving us accurate clipping indications?

Strange.

Stephen

 

[Stuart English]

Build 30 meters

Build 30 meters

Would need to see a JPEG of your monitoring c/w the red EXPOSURE overlay.

 

[Julio Quintana]

I also find my self consistently confused by the relationship between the histogram, stop lights, and little colored bar. I regularly get the red stop light when my histogram clearly isn't touching the right goal post. I used to always set the zebra stripes at 105 and knew that anything above that was clipping (practically). Now the zebras don't seem to work the way they used to (or like every other video camera I have ever used). It would be nice if someone at RED would post a short video on the 1-2-3's of exposing for Build 30. I know Jim has posted several times on the issue, but I would assume that most people on this forum are visual learners .

 

[John Brawley]

I also find my self consistently confused by the relationship between the histogram, stop lights, and little colored bar. I regularly get the red stop light when my histogram clearly isn't touching the right goal post. I used to always set the zebra stripes at 105 and knew that anything above that was clipping (practically). Now the zebras don't seem to work the way they used to (or like every other video camera I have ever used). It would be nice if someone at RED would post a short video on the 1-2-3's of exposing for Build 30. I know Jim has posted several times on the issue, but I would assume that most people on this forum are visual learners .

A video won't explain it. Zebras won't work the way were used to them working anymore unless they change it back...

Look here

http://reduser.net/forum/showthread.php?t=44812

 

[Stephen Pruitt]

Hey there Stuart. . . I don't use or like the monitoring exposure overlay. I want to see the image in color all of the time without overlays.

I just want to know which meter to use. . . the RGB levels on the graph? The goal post? The stoplight? The exposure bars?

I would think they would be able to give me the same information, but they clearly are not. The stoplight and the exposure bars appear to be pretty consistent with each other, but they yield very different information than the goal posts. That was a genius addition to the system, but it apparently doesn't tell me what is really in noise or clipping. What's the point of the meter in that case?

Stephen

 

[Julio Quintana]

A video won't explain it. Zebras won't work the way were used to them working anymore unless they change it back...

Look here

http://reduser.net/forum/showthread.php?t=44812

So what exactly does the zebra function do, because it used to be my most useful tool for guarding against clipping, and now it seems useless.

 

[Sam Winzar]

The traffic lights and the the little RAW pole will show clipping when just a small amount of the image is clipping (do I remember Stuart saying 2% in a post?). The poll to the right of the histogram gets larger as more of the image clips. Could it be that you have a tiny amount clipping which is enough to trigger the traffic lights but barely reads on the histogram pole?

The easy way to see would be to use the RAW EXPOSURE overlay as it will show you exactly where and how much of the image is clipping. I know the overlay isn't to everyone's taste but I think it's a good tool that you can at least use to better understand the other exposure tools by showing what they are reacting to.

 

[Julio Quintana]

Not to beat a dead horse, but can someone please explain to me why the we can't have normal zebra stripes that appear when viewing in Redcolor but reflect the true exposure of the Raw image? Tons of ENG and documentary guys work that way and I don't understand why it can't work in addition to the other exposure tools.

Even if I am shooting M-X at ISO 2000, it would be great if I could use the zebras to tell me exactly what part of the image is actually clipping in the Raw file.

 

[Stuart English]

Zebras etc

Zebras etc

Because Zebras work on the Luma content of the RGB monitored image - just like on all other cameras. And the conclusion of the thread linked to above is that we disagree on the premise that Zebras functionality changed between Build 21 and 30.

Are you actually saying that Zebras are indicating that you are NOT clipped when in fact you are?


As for a traffic light triggering, it is as described. The vertical bar is a relative indication of the AREA of the image that is in clip, the traffic light is an absolute indication that SOMEWHERE in the image is clipping.

 

[Sam Winzar]

Julio, I understand exactly what you are getting at but I don't think you understand Stuarts point of view.

Zebras on the RED One do the EXACT same thing that ENG cameras do. They give an indication of where exposure is at based on the image that has been processed by the camera. You wouldn't expect an ENG camera that you are shooting with +9db gain to have it's Zebras react to the 0db gain image?

As the RED One is not an ENG style camera ENG style exposure assists aren't going to suit it. That's why Red have given us the exposure check but I can understand the desire for RAW Zebras as an option.

EDIT *well that just got summed up by Stuart pretty much as I was posting this*

 

[John Brawley]

Julio, I understand exactly what you are getting at but I don't think you understand Stuarts point of view.

Zebras on the RED One do the EXACT same thing that ENG cameras do.

They do not do the same thing.

Set your RED to 320, set your zebra to 102-106 while monitoring in redcolour and point it at a source that clips the sensor and tell me if you see any zebras....

You wont because they roll off super white or 100+ values now with the new lut.

Every other zebra DOES allow you to set a zebra that is in the 100+ range.

RED's position is that these values aren't useable and that 99 is the new 108. Well actually, values from 100-108 are represented by 99. Confusingly though, 98 is still 98. The SCALE has changed.

It's just that RED still allows you to set a zebra at 100+ values....it just doesn't actually WORK.

This is why this is such an idiotic decision. RED should re-instate REC709 monitoring as an option and then allow the zebras (and FC for that matter) to behave as they once did. Let us CHOOSE which LUT we monitor and subsequently expose with.

That's why Red have us the exposure check but I can understand the desire for RAW Zebras as an option.

Exposure check doesn't work while you're in the middle of a take unless you're happy for the director, client, operators to be watching a black and white image with the occasions red sizzle.

This issue keeps on coming up and it's not useful to keep denying that it's an issue and saying there are other tools that do the job better. They don't and zebra isn't the same as every other camera.

jb

 

[Stephen Pruitt]

Stuart. . .

. . . unless you are telling me that the goal post and histogram are really inaccurate tools, what you are saying is not correct. . . at least not on my camera.

I can get the stop light to light up completely and the color bars to show maximum exposure and not even have the histogram touch the goal post, let alone register on the goal post as clipping.

I haven't done the calculations, but that goalpost has got to be at least 50 pixels tall. If that were the case, then, if it were properly calibrated, when the stop lights are lit, showing a minimum of 2% of the pixels in the image are blown, I should see a line, at least one pixel tall, lit up on the goalpost. I don't. I apparently have to be pretty far into clipping before anything on the goalpost lights. If that's the case, then what on earth good is the goalpost?

Is the same true on the low end? That is, am I actually into noise when the lower goalpost isn't showing action?

I had originally thought that the goalposts were a god-send. Now, I'm not sure if they are really telling me anything or not. Does the overlay tool tell me accurately where I'm in noise? I'm much more concerned about noise than I am clipping.

Stephen

 

[Julio Quintana]

RED's position is that these values aren't useable and that 99 is the new 108. Well actually, values from 100-108 are represented by 99. Confusingly though, 98 is still 98. The SCALE has changed.

jb

I want to monitor in Redcolor and have my monitor show zebra stripes over whatever part of the frame is clipping in the raw image. I don't want the image in black and white, or "Predator vision", as it's extremely distracting during a take. Would it work that way if I set the zebras to 98-108?

 

[Stuart English]

Exposure Tools

Exposure Tools

Stuart. . .
. . unless you are telling me that the goal post and histogram are really inaccurate tools, what you are saying is not correct. . . at least not on my camera.

I want you then to record a 1 frame .R3D and send that to me along with a photo of the UI screen when you see this issue.

stuart@red.com

 

[Stuart English]

Zebra and sensor clip

Zebra and sensor clip

I want to monitor in Redcolor and have my monitor show zebra stripes over whatever part of the frame is clipping in the raw image.

Yes, just set your Zebra to 99, which is why it is the default value.

 

[Julio Quintana]

You wouldn't expect an ENG camera that you are shooting with +9db gain to have it's Zebras react to the 0db gain image?

I would if the ENG camera was actually recording at 0db but just making the monitor look like +9db, which is what the RED is doing.

So Stuart, are you saying that it's just not possible to monitor in Redcolor but have zebras that reflect the raw image? Is there an actual hardware limitation that prevents this?

 

[Julio Quintana]

Yes, just set your Zebra to 99, which is why it is the default value.

Oops, just saw your post. So setting the Zebra to 99 will show actual clipping in the raw image even if I'm watching Redcolor?

 

[Stephen Pruitt]

Stuart. . .

I'll get that shot of the monitor sent to you (and posted here) later today. I need a digital camera with a macro capability, and all I have is a cell phone at the moment, but I have reproduced the issue upstairs right now.

The histogram is not only not touching the goalpost, but it is like a goalpost width away from the goalpost. (I am blasting an HMI at a white wall so I have a rock wall drop off.) And yet, the lightbar clearly shows top exposure and the stoplight clearly shows 2% of the pixels (green, in this case) at clipping. But there is no red (noise) shown on the goalpost at all.

I do not believe I will be able to send you a single frame, as my email limit is 20 meg.

Stephen

 

[Stuart English]

Zebras etc

Zebras etc

So setting the Zebra to 99 will show actual clipping in the raw image even if I'm watching Redcolor?

No, Zebras are there to show you video levels of the HD-SDI output.

 

[Stuart English]

Single Frame

Single Frame

I do not believe I will be able to send you a single frame, as my email limit is 20 meg.

Give it a try, a single frame .R3D is quite small.. or shoot at 2K and RC28.