How to avoid this in render?

When I render an animation with dark lights or very bright lights, it always has these choppy color gradients. When you look closefully at the colors you can see that it has not a constant color gradient. But when Im in blender, it doesnt have it, only the render version.
How can I avoid this in render, which settings do I need?
I only noticed that when I use ,constant bitrate" that its less strong, but its still there.


When you say “constant bit rate” does that mean you are rendering to a video? If so, that could be the issue. You may be experiencing loss of information due to the video compression. It is almost always advisable to render to still images first and then to use those images to render your video - especially if quality is important. Could you post the settings you used to render to video?


yeah like Hunka said- atleast render in png sequence. Tiff is way better.

1 Like

So i shouldnt render it as mpeg4´?

yes, dont render mpeg4. Select png- then it will render a png sequence. After that u can composite in blender or any editing software like hitlfim/ after effects/ premiere

I did it as png and added it to blender video editor. I added it as image sequence and output was mpeg4, and it still looks the same and not good

so is the png sequence render any better than mpeg4 u exported? or the problem is still visible in png sequence?

And whats "“not good?”

the problem is not visible in png sequence, only when i render it as video again. The quality is still bad (the colors are still like in the images)

Then try one of these Hitfilm / after effects/ premiere- and export in .mp4.

I use Hitfilm because its free : )

ok, i will try but here i want to show you what i mean
this is how it looks as a video and i dont like the quality:

and this is how it looks as png image in a very good quality

you can clearly see the difference when you look at the wall

Yes, that compression is really bad- thats why i never used anything other than png in my work.

1 Like

Without being an FFMPEG guru, what I interpret the Constant Bitrate setting in conjunction with the Bitrate, Minimum, and Maximum settings is that every frame will be capped at 6,000 - regardless of the 9,000 maximum setting. What that means is that for frames that need more bits to represent more gradations of color as it transitions in a gradient from one shade to another, you will have banding because there aren’t enough bits. You can either:

  1. raise your constant bitrate value - this will also raise your file size because every frame now has more bits to use to describe the image, or
  2. switch to a variable bitrate utilizing the Minimum and Maximum values that ensure you have enough bits to keep from banding - this may take a bit of trial and error with some pixel peeping, but you stand to have a smaller file when you are done.

Let us know what you decide to do and share your results if you can! :slight_smile:

1 Like

@123ERTYY, not to derail the thread too much, but what would you say makes TIFF better than PNG?

With the other software it looks much better! But I thought its free? Are you sure its free?

I increased the bitrates from 6k to 100k and im happy with the results now, thanks for your help!! :slight_smile:


and here in close

Do you know what the maximum of bitrates i can use is?

I think you have used an add on thats not free : P in the effects panel try to get the ones without that green “Add-on” box.


1 Like

So, I couldn’t find a super straight answer as most results were about how specific encoding libraries encode to the H.264 standard. I did find this in a page for a product that uses the x264 encoding library just like FFMPEG:

The bitrate chosen is 20 Mbps, which is a safe 80% of the maximum peak bitrate allowed for H.264 level 4.0.

So, I believe in the H.264 standard, there are different profiles that have different associated encoding flags, and I believe 4.0 is the “Main” profile that most decoders aim to be compatible with. Even if you maxed out the bitrate (100% of the Main profile max), you then would require a pretty beefy machine to decode all those bits. So, at least that gives you an idea of where your upper bound should lie.

Here’s the site if you want to read some more:

1 Like