Histogram not tightly enclosed by raw-bars as in manual

[Alvise Tedesco]

I need to be higher than 800 to see more than yellow and no gap..

 

[Carl Dieker]

Only being able to get the histogram working at different ISO:s sort takes the whole point out of getting proper exposure. To fully take use of the histogram and the Videotool the image has to be set to RAW.

 

[Carl Dieker]

Different builds have shown, at least on my Scarlet, slightly different gaps (still no more then yellow on false color tool).
It looks to me like it would be possible to trim it correctly via firmware, but it should be an individual one (single camera specific). And that could be the reason we haven't seen plain answer from Red?

An interesting but not comforting idea...

 

[Thomas Koch]

I have had a few ruined shots because of this bug. UGH. I guess you shouldn't read the manual and follow it....

 

[JanneJansson]

Is this a bug or a hardware problem? I have had this on my EPIC-X #756 from first day.

 

[Carl Dieker]

I would like to know if everyone are affected by this. Even if it might be so that the Epic is built to handle the histogram like many of us have complained about, it would be very helpful to get the R1 handlingen of the histogram back as an option. These cameras are all about exposing as correctly, creatively and intuitely as posible and then just pressing record. Filming in REDCODE made all the special pre-defined and definite looks, gamma levels, WB and ISO obsolete. The way false color/video mode used to work was that it showed both in RAW and RGB very directly how the whole piture was exposed in different and multiple areas/levels: purple, blue, grey, green, pink, yellow, orange and red. And this, as already mentioned, worked both in RAW display as well as in RC2-3, Rg2-3. The exposure mode and the zebras only shows two modes of exposure, underexposed or/and overexposed. Also the R1 RAW/RGB histogram went all the to the right, no matter if RGB displayed the cameras potential range of highlights or not. Amen.

 

[Carl Dieker]

I just got an email from Matt Biederman at RED that gives me hope for the future about the possible development of new exposure tools. It would appear that the gap in the histogram and lack of yellow, orange and red in Video mode is nor a hardware or a software problem... Thanks RED for listening.

 

[Patrick Grossien]

Alright, trying to get this right, so this one will be long.
I would love to know if this is what RED intended it to be like.
Ok, here goes:

The histogram is showing us metadata reference, depending on what look is applied.
When you switch to RAW check view, it shows RAW reference.
Below are some UI screenshots I took to compare that.

The first is an overall well exposed shot with different ISO settings. Look at what the Histogramm is doing.
The camera is rock steady, Image is not moving.
When the look is applied @ ISO 12800, it shows clipping in the histogram but the RAW data is still well exposed
(check the bar and the traffic lights).

1a. well exposed image @ ISO 250 / RC3, RG3

1b. same image @ ISO 250 / RAW VIEW

1c. same image @ ISO 12800 / RC3, RG3

1d. same image @ ISO 12800 / RAW VIEW

OK, now if I shoot a different, clipping image, and compare the histogram I get the results below.
When switched to RAW view, the histogram is tightly enclosed at the right, but not "hitting the ceiling".
Since I'm not used to this kind of representing the image values, I created a fake histogram, to what I assume it would "normally" look like.
So picture 2c which is what I believe everyone is looking for: a "hitting the ceiling" histogram even with metadata look applied.


2a. clipped image @ ISO 280, f 2.8 / RC3, RG3 (with gap)

2b. clipped image @ ISO 280, f 2.8 / RAW (no gap, but not "hitting the ceiling")

2c. clipped image @ ISO 280, f.28 / RC3, RG3 FAKE HISTOGRAM!

And last but not least, the histogram below is what I believe is creating the problems for most who are addressing this.
The image is not clipping by a lot, but it is as you can see by the clipping bar and the traffic lights.
And because the histogram looks like a well exposed image, no one who is used to only using histograms
to check on the highlights, would think this is image is really clipping.


3. different image (with little clipping) @ ISO 280, f16 / RC3, RG3

So here's where the traffic lights come in. As long as you check these, and maybe switch to RAW check view
if you are unsure, you should be fine. I don't rely on the histogram to check clipping but only on the right bar and the traffic lights.
If this is not a bug, but just a different way to go at it, I still believe that there could be a way to make the histogram represent
what everyone was used to working with (as in img 2c) Or maybe this just needs to be explained "as if we were 4 years old"

BTW this is represented the same way in RCX. The waveforms stop at around 80-90% also.
The only time the histogram ends up all the way to the right in this image 3, is with activated RedLog. And RedLogFilm shows the gap, too.

 

[Carl Dieker]

Interesting post Patrick. Me too don't rely on the histogram but base my info on hightlights clipping primary on the traffic lights. However, when taking a fast glimpse at the histogram and seeing that it hits the ceiling make me watch the traffic lights more carefully and more often. This is a habit picked up from shooting with R1 for 3,5 yrs. I like that habit.

 

[Patrick Grossien]

Carl, I have been used to relying on Zebra and Histograms with most cameras I've used in the last 10-20 years. So this is baked in into me But since my scarlet, I've been ignoring it mostly, which I find kinda sad, because with a fast glimpse I knew what was going on. So I think I would also like to have it the way I was used to, but not entirely sure, because as I understood it, the whole thing is intended. And I guess we're just not educated into why this is the way it is. What kind of advantage it holds, etc.

I guess basically what I'm saying is:
If this is not intended I'll still be a somewhat happy camper until they find a fix, because until then I can accomodate with bar and lights.
If there is something going on with the sensor not capturing full bandwidth, please let me know asap.
If it IS intended I'd love to gain more understanding on the whole subject. So please do REDucate us, guys

 

[Patrick Grossien]

RAW view is essentially setting ISO to 320 in whatever curve you are viewing. If you compare RAW in RLF vs RG3 you'll see what I'm talking about. But bottom line is that the histogram represents a purely metadata view of the data.

There are two ways of representing this..

1) show the full width of the histogram which represents every possible bin in any mode (this is what we're doing now) .. but right now once you go over the max value all values go to 0 but the raw clip meters and stop lights indicate clipping.

2) clamp the top of the histogram to the maximum achievable value and show a stretched view of the histogram. (Portraying things in this way may alleviate some of the confusion, but will not be consistent with RCX)

This is what I meant in my statement about the whole thing being intended. Maybe Matt, you could educate us how to read the histogram the way it is now. Because I can't quite grasp it.
Best explained with pictures, examples, descriptions, arrows and of course cute little animals.

 

[ericyoung]

RAW view is essentially setting ISO to 320 in whatever curve you are viewing. If you compare RAW in RLF vs RG3 you'll see what I'm talking about. But bottom line is that the histogram represents a purely metadata view of the data.

There are two ways of representing this..

1) show the full width of the histogram which represents every possible bin in any mode (this is what we're doing now) .. but right now once you go over the max value all values go to 0 but the raw clip meters and stop lights indicate clipping.

2) clamp the top of the histogram to the maximum achievable value and show a stretched view of the histogram. (Portraying things in this way may alleviate some of the confusion, but will not be consistent with RCX)

I think this thread is really a debate on whether a technically "correct" tool is intuitively useful (similarly with the post FLUT False Color view mode). I mainly use the RAW Exposure View mode, and RAW goalpost/stop lights in preference for this reason.

For the purposes of image acquisition, I think the second "stretched" histogram mode would be more intuitive and useful to camera crew accustomed to using histograms on non-RED cameras, and perhaps should be the default histogram mode. For those who are more technically minded or who like to read the manual or really get to know the camera or do their own RCX tweaking, then the first "full range" histogram mode could be a selectable option. Of course there should then be a very clear indication in the displays of which mode is activated!! Also a very bold clear note in the Release History notes so that we know there's been a change in monitoring tools.

I'd also like the display indication for RAW view mode to be clearer than the word "OFF" in different colours. Didn't it used to change to "R" with a tick once upon a time? Maybe "RAW" and "COL" (or "META") would be clearer, although the ticked "R" would be fine too.

On the related False Color view mode, I'd like a default option of red areas showing where FLUT is beginning to work, so I'd know where highlight areas are getting rolled off/compressed (this might need a indication threshold control). Purple where shadows are getting crushed too. Basically similar to the RAW Exposure view mode, but for metadata.

 

[Patrick Grossien]

Eric, good post! +1 on your statements.

 

[Petri Teittinen]

It would appear that the gap in the histogram and lack of yellow, orange and red in Video mode is nor a hardware or a software problem.

How can it be neither of those? What else is there besides hardware and software? Is the bug caused by weather?

 

[Carl Dieker]

How can it be neither of those? What else is there besides hardware and software? Is the bug caused by weather?

Yeah, I believe what Matt Biederman/RED says it's not a soft/hard bug but intended as commented above. Myself, I'm still not sure if it is a bug or not... REDucation, please!

 

[Carl Dieker]

Sounds good, Eric.

 

[Patrick Grossien]

How can it be neither of those? What else is there besides hardware and software? Is the bug caused by weather?

It could be intended as mentioned above.

 

[TheBied]

For the purposes of image acquisition, I think the second "stretched" histogram mode would be more intuitive and useful to camera crew accustomed to using histograms on non-RED cameras, and perhaps should be the default histogram mode. For those who are more technically minded or who like to read the manual or really get to know the camera or do their own RCX tweaking, then the first "full range" histogram mode could be a selectable option. Of course there should then be a very clear indication in the displays of which mode is activated!! Also a very bold clear note in the Release History notes so that we know there's been a change in monitoring tools.

Adding a "stretched" mode is tricky since R, G and B can have different clip points. Stretching per color will lead to different scales per color and a skewed color balance shown in the histogram.

The histogram as it stands shows you what the color science is actually doing to the data. Just as you can view the histogram in RCX while tweaking any parameter, the camera offers the same flexibility.

We are exploring ways of making everything much more intuitive, so there will be improvements.

I'd also like the display indication for RAW view mode to be clearer than the word "OFF" in different colours. Didn't it used to change to "R" with a tick once upon a time? Maybe "RAW" and "COL" (or "META") would be clearer, although the ticked "R" would be fine too.

Completely agree. Currently, we're leaning towards removing the colored "OFF" from the status pane and showing clearly in the histogram itself which look parameters are being used. So you'd see an overlay in the histogram stating RAW, RC3/RG3, etc.

 

[Petri Teittinen]

It could be intended as mentioned above.

Ah, the classic "It's not a bug, it's a feature" solution, eh?

 

[Kevin Marshall]

Yeah, I don't think this is a bug that affects only some cameras. I've seen it on each of the 4 Epics and 5 Scarlets I've worked on, from build 1.7 up till 3.2.16. It is a little disconcerting to have a fully black or fully clipped image at 800 ISO slam to points not quite fully over to the left/right, especially since the histogram is ISO-dependant. But for what it's worth, my EX1 does the same thing with its histogram.

I do agree, though, that it would be helpful for the end points of the histogram to be the goalposts at all ISOs.