I don't use 32-bit precision because of the internet, just don't want to have problems using a chain of several plugins.

Using 8-bit is not a solution to the problem, it's a workaround :)

I didn't say "because" of the internet; you have been fooled into alchemic thinking by misinformation you read on the internet or elsewhere.

8 bit media goes in an 8 bit project; anything else is asking for processing problems that are well documented here.

If you pour 8 gallons of water in a 32 gallon bucket, it is still 8 gallons.

The notion of processing effects in 32 bit space and then dithering down to 8 bit for production is an illusion, because doing so creates frequency domain noise in the lower two bits that more than negates any theoretical advantage with regard to precision. Only when you have 10 bit source is there any advantage. We've seen that rodeo.

Welcome to the peer support forum for Vegas Pro.

Does using curves in the Color Grading Panel work better than the curves Fx when in 32-bit mode?

Does using curves in the Color Grading Panel work better than the curves Fx when in 32-bit mode?

Little. The clip does not flicker (at least in the preview).
But only part of the graph (far left side) works and covers the whole range.
Just do as in the screenshot to get a completely dark image.

I have to check it out, because it probably shouldn't be like that.

For now, I checked the usefulness of 32-bits by adding the Contrast & Brightness plugin with the Brightness +1.0 parameter, and then the same with Brightness -1.0. I got the original back.
With 8-bit, this is not possible at all.
With 32-bit I can (almost) anything :)

Suit yourself 😉

Does using curves in the Color Grading Panel work better than the curves Fx when in 32-bit mode?

Little. The clip does not flicker (at least in the preview).
But only part of the graph (far left side) works and covers the whole range.
Just do as in the screenshot to get a completely dark image.

I'm perplexed by this behavior. What type of media is this and can you share a sample? Are you using the color spaces in 32-bit mode (32-bit full)?

I don't see it myself. Here's 32-bit video (after a levels correction of studio to computer RGB done at the media level)

Here's 32-bit full with view transform off.

My file is 13GB big :)
I'll try to find something smaller or cut out a part.

Have you tried with the Color Curves?

Can you just record a few seconds with the same settings (if it's from a camera)?

Here's 32-bit full, view transform off and color curves Fx on the event:

I prefer 32-bit video as it's more predictable. Generally though I use 8-bit full as my footage is 8-bit.

Try this clip. It's recompressed in Vegas but the problem still persists. I am getting random image flickering.

And in version 18 and in version 19.

https://files.fm/u/e6hx9mh5j

(set Preview quality to Best(Full), compositing gamma must be set to 1.00 (Linear). With the setting 2.22 (Video), the error appears very rarely)

1, that's not flickering, it's low light noise.

2, fake declaration: your clip has a brightness/contrast gradient.

The fist frame

And the last one

Solution:
Filter Brightness&Contrast->Automatic Contrast->take last frame as current referrence to calculate the B&C.
Then change to set for each frame independently and calculate the B&C.

For now, I checked the usefulness of 32-bits by adding the Contrast & Brightness plugin with the Brightness +1.0 parameter, and then the same with Brightness -1.0. I got the original back.

Of course not. Redundant stacking of levels filters reduces bit depth, always. You are creating an optical illusion, and your controlled testing will prove this. Many people prefer nice-sounding theories over demonstrable fact. The word scientists use is "confabulation." Carry on.

1, that's not flickering, it's low light noise.

Google Photos massacred the quality, but the problem is clearly visible in 2 frames.

https://files.fm/u/c8dg7dfhd (6s i 12s)

For now, I checked the usefulness of 32-bits by adding the Contrast & Brightness plugin with the Brightness +1.0 parameter, and then the same with Brightness -1.0. I got the original back.

Of course not. Redundant stacking of levels filters reduces bit depth, always. You are creating an optical illusion, and your controlled testing will prove this. Many people prefer nice-sounding theories over demonstrable fact. The word scientists use is "confabulation." Carry on.

Show me how you do something like that in 8-bits and I will switch immediately :)
And why on Waveform and Histogram after such reverse operations there is no difference?

You are asking me to demonstrate something that is physically impossible? I'll leave that part to internet soothsayers.

But if you want to design a controlled testing environment using tools such as RMS regression, SSIM, and VMAF mediated by PSNR, we like that here. That's not a trivial suggestion, because my own tests could use updating.

Here is a picture that magnifies the frequency domain noise that populates the lower 2 bits when switching environments such as you propose. This does not even take into account the likelihood of collateral gamma shifts, dither artifacts and quantization noise in the spatial domain. The other artifacts you report may be due to your graphics drivers, but I'm really not interested in pursuing that because it is secondary.

By all means, keep exploring! Peace and learning.

;?)