compositor nodes and image render size

Hi everyone,

it might be a stupid or just a too simple question, but I couldn’t find a solution, so I have to ask it. :o
I used a composition nodes setup and got a nice result. Then I changed the render size of the image from 800x600 (first image in attachment) to 1280x960 (second image in attachment) and it doesn’t look the same.

Is there a way to change node settings related to the render size of the image?
I can group all nodes, but is there e.g. a node for changing the values regarding to the image render size or do I have to adjust it node by node on my own?

Thanks in advance!


Add>Distort>Scale. All nodes calculations happen on a linear, temporal fashion, there are no global settings other than the currently rendered scene’s settings. Output size is always based on your camera/scene render settings.

Hi RamboBaby, thank you for your answer!
Yes, I tried the scale node, but it just scales the image segment. As you said, the image render size is set int the render settings.
I tried a heavy use of Convertor->Math-Node with Multiply and Add to the NodeGroup. It helped a bit, but I had to adjust a lot by myself and added also addtional Nodes to get a similar result.

So there is no other way than to set the Composition Nodes in the final render size?
(As changing the render size, will change the result as well.)
Or is there another way to adjust Node values?

BTW, that’s the image, that I get with 4 Math-Nodes (multiplied by 1.6, as the image size is enlarged by 1.6) and an additional contrast adjusting node.

It’s still not the same, so I appreciate any help with that. :slight_smile:


use the scale node, 100% relative, iirc

PapaSmurf, please forgive me, I do not understand (as I’m not very experienced).
I tried to use the scale node, positioned before and/or after the Node-Group and tried also different values for Relative and SceneSize%. It just scales the image, but doesn’t affect the node effects.
Using a blur node with Fast Gauss and y=32 on an image with 800x600 doesn’t have the same effect as it has on an image with 1280x960, does it?
I also tried unsuccessfully to use scale “relative 0.625” before the node group and scale after the node group with “relative 1.6” with a render size of 1280x960 (to simulate a smaller image size, before applying node effects).
There is no way to connect changes of render size to values/settings of composition nodes automatically to achieve the same result?

I 'm fairly sure this has nothing to do with what your nodes are calculating and is more than likely the result camera having nothing to actually focus on (geometry/depth buffer). It’s grabbing a slightly different perspective of the halos and the way that they blend with each other every time the scene size increases. I think it’s more of an interpreted result rather than an absolute result since the halos are phantom objects.

Ah, that explains it fairly well. Thank you very much RamboBaby!
But (sorry for being that annoying) that means, it only shows this behaviour with halo-material? And I would get pretty same images while using a “solid” material and changing the render size?
It’s not a good example, as it is a quick setup, but it slightly shows the difference. Rendered in 800x600, the “glow” is wider than in 1280x960.
So like in the halo-scene, I had to adjust the composition to the render size change or better create the composition nodes for the final render size?
Or what else can I do?

Sorry, deleted double post.