Could you share a link to download the source video?

Photoshop Histogram of your photo

I used a Levels filter in Photoshop to clip all three channels of your photo.

Photoshop Histogram of your photo with Levels filter applied

Clipped in Vegas Histogram

Clipped in Vegas Waveform

Clipped in Vegas RGB Parade

Clipped in Vegas Vectorscope

Here is a screenshot of photoshop directly ingesting the source video:

All three channels are clipping:

Here is Resolve (showing clipping in both histogram and waveform):

(edit here) OK, photoshop and resolve are levels-correcting the original video clip to 16-235; I'm using Vegas in legacy 8-bit mode so Vegas is not altering the levels.

If you click the image in my original post and save it locally, the levels match the source video, so I don't think there's any reason to download the full video.

In the second image in my original post, Vegas' histogram is clearly showing highlight clipping in all three channels; the waveform and rgb parade are not. This is in 8-bit legacy mode, with the studio levels box unchecked in scopes.

As far as I know, the histogram and RGB parade always operate in 0-255, and with the box unchecked, waveform is showing 0-255 also. So why don't they match?

What the scopes are showing if you switch from "Legacy 8-bit (video levels)" to "8-bit (full range)" in Project Properties?

@ALO

With all due respect, it seems we have a vocabulary malfunction.

On your version of the Photoshop histogram, only the red chroma is bumped ever-so-slightly against 255 . Totally expected and normal. We never label an image as being "clipped" based on a single chroma channel, so as not to make a blanket indictment of 98% of the outdoor footage ever shot 😶

There is no evidence of Y' (luminance) clipping in your histogram, therefore your definition of clipping is nonstandard. @john_dennis showed you examples of deliberately clipped footage. I won't repeat the advice that you spend a lot of time observing a lot of camera and post production scopes before making attributions. It took me exactly 2.7125 years of training to understand waveforms and color histograms, and I started at Technicolor (yes, that one).

Maybe this informative and humorous look at the whole levels conundrum will help us get on the same bus with regards to nomenclature, as several examples of clipped and not-clipped footage are illustrated.

https://www.vegascreativesoftware.info/us/forum/pc-to-tv-levels-a-comedy-of-errors--107325/

It would be helpful to see the original file as PNGs are data levels (0-255). Not sure what the source footage is. In Resolve can you right click on the media and set the video as a full range file? It might have superwhites that are being clipped.

None of the scopes in Vegas are all that high-resolution but with your image in VP 18 I see some data values near 100% for both RGB Parade and the waveform. (I trust the histogram the least out of the scopes in Vegas)

@ALO

With all due respect, it seems we have a vocabulary malfunction.

On your version of the Photoshop histogram, only the red chroma is bumped ever-so-slightly against 255 .

To me, Vegas' histogram is indicating clipping in the combined luma and individual channel graphs because the values stack up on the y-axis at the far right (255) line in each graph.

But if we bring the video image into Photoshop by importing a png (to prevent PS from automatically converting the levels) we get this:

Which indicates essentially no clipping (we agree on that!). If I didn't know better, I would think we are looking at histograms for two different images. What do you see? How are you interpreting Vegas' histograms?

OK, suit yourself.

But I think what you are misinterpreting is the difference between the linear histogram in Photoshop, with which you are familiar, and the logarithmic histogram in Vegas, with which you are probably unfamiliar. Sorry for not making that distinction clear sooner.

Both are correct.

BTW, I have previously made requests for a switchable linear /log histo in Vegas Pro.

The Photoshop histogram you show above seems to be linear. The Vegas Pro histogram is logarithmic and can be very misleading because same luma/RGB level of only a few pixels could have a big impact onto the graph.

If you have a tool which lets switch the histogram from linear to logarithmic (KdenLive does) then one version would be close to Vegas Pro (it's still likely they would use a different logarithmic math) and the other one would be close to Photoshop.

This is your frame viewed in the KdenLive histogram with these two options:

Linear

Logarithmic

Same frame, same tool, same histogram but different math base (linear vs. logarithmic).

And neither are clipped.

@Marco. Thank you!

The Vegas Pro histogram is logarithmic and can be very misleading because same luma/RGB level of only a few pixels could have a big impact onto the graph.

I had no idea. Thank you -- that explains a lot!

Another example for how misleading the histogram may be.

Take a close look at the histogram bars below. There are two on the first screenshot and three on the second.

Now guess what the difference is in the corresponding frame to the right. You don't see it? No?

In the lower screenshot there are a couple of small white dots inside the brighter square. You would never ever notice them in a real recording but the histogram does as if there's something which might need to get corrected.

Here is a VP18 demo project if someone wants to have some fun to repro. If you do, be sure to have the preview set to "Best (Full)" and the width of the histogram made large enough.

Yikes! I'm confused by the log scale. Reading online it looks like log histograms use different-width bands for the x-axis, but as far as I can tell (including from Vegas' numerical indicator) all the bands on the x-axis are equal-width. Is this a log scale in the y-axis? How do you interpret Vegas' histogram -- and why would you want the data scaled this way?

See this curve which represents the math used for the log graph of the histogram:

The XY axis values 1 mean 100 %, so 100 % of the project's pixel dimension for X and 100 % of the histogram height for Y.

You see if 50 % of all the pixels of your frame has exactly same luma/RGB level, its histogram bar would have a height of more than 90 %.

Or if a histogram bar is 50 % of the maximum height this represents less then 2 % of the project's pixel amount.

"Why would you want the data scaled this way?"

Not sure why. I am very careful using the VP histogram and I tend to call it a forensic tool. Nice to analyze noise, swings caused by sharpening/aliasing, find banding and so on. Some limited use to evaluate the balance of contrast/gamma, though that's where it starts getting misleading.

On the other hand a pure linear scale would make it hard to see when a smaller region or a amount of pixels – which still could be big or much enough to matter – is out of a targeted levels range. Remember: less than 2 % of the project's pixel amount would plot a histogram bar of 50 % height in VP. But 2 % of an HD frame are still around 40000 pixels. A logarithmic scale is fine, it just must not be that harsh as it is in VP.

Now here is another approach to see how the histogram works. If you want to have a more or less linear progress of the bars in the histogram like seen here

You need to use a graysale like this one (click onto it to enlarge)

The upper numbers are the amount of pixels, the lower numbers are the RGB/luma levels of these pixels. So in this case for the very left histogram bar there are only 32 pixels with 0 % luma (RGB 0) and for the very right bar there are 8192 pixels with 100 % luma (RGB 255).

Yes, precisely. Here's a side-by-side curve comparison. Note that linear display favors amplitude, log display favors density.

"Why would you want the data scaled this way?"

Same reason that camera aperture and exposure values have been scaled this way for163 years. Analog PMT voltages soar on a linear scale. Twice the light, ten times the encergy. It's good to learn to live without the exaggerated appearance.

Looks like Uncle Ansel's been waiting for you:

https://www.amazon.com/gp/product/B078GKPH5G/ref=dbs_a_def_rwt_hsch_vapi_taft_p1_i0

I find the Vegas histogram okay for a quick check of black and white points though it may mislead with a little bit of data appearing to be more significant than it really is.

For what I do it's not a very useful tool- I like to see where my shadows or skies are so I can target them with curves for example.

With Vegas histogram they are just lumped together in an indeterminate mass. With Graide Color Curves I can use the eyedropper to find the values I want so it's okay not to have a histogram. With regular Vegas curves I really need to use a different tool, like the waveform monitor, to see what I am doing.

.

The Vegas histogram approach is extreme in my opinion and would hope they revisit it at some point. It would be nice to have a histogram underneath the color curve in the color grading panel, too.

I find the Vegas histogram okay for a quick check of black and white points though it may mislead with a little bit of data appearing to be more significant than it really is.

For what I do it's not a very useful tool- I like to see where my shadows or skies are so I can target them with curves for example.

With Vegas histogram they are just lumped together in an indeterminate mass. With Graide Color Curves I can use the eyedropper to find the values I want so it's okay not to have a histogram. With regular Vegas curves I really need to use a different tool, like the waveform monitor, to see what I am doing.

.

The Vegas histogram approach is extreme in my opinion and would hope they revisit it at some point. It would be nice to have a histogram underneath the color curve in the color grading panel, too.

+1 on that. By the way thanks for the Leeming LUT tip. I've been playing around with his GoPro correction and learning a lot about how to pull more data out of the flat/native profile.

Musicvid, I am sorry to tell you if Ansel were alive today he'd be making TikTok videos and asking for a non-weird histogram. :)

If Ansel were around today his assistants would be pioneering techniques for HDR digital capture and projection. They'd be leveraging AI and computational photography in their craft towards the end of helping protect the last natural places by helping us feel the awe of standing before them. : )

Glad the Leeming LUT tip helped.

Musicvid, I am sorry to tell you if Ansel were alive today he'd be making TikTok videos and asking for a non-weird histogram. :)

If Ansel were around today his assistants would be pioneering techniques for HDR digital capture and projection.

No, he would still be teaching the basics, which most digital-era pretenders have never bothered to learn.

Lacking that, it's a good thing modern-day cameras do everything automatically, so the uneducated will have something to blame.

There are no clipped levels in the shot you uploaded.

well, at least in your screenshot blue is clipping.

[sigh] I givie up. "Clipping" can be any old thing you want it to be. Cheers.

"well, at least in your screenshot blue is clipping."

It's not. The height of a histogram curve represents the amount of pixels, not the level.