Copy link to clipboard
Copied
I'm sorry if this is a stupid question, but I'm not quite understanding how compression of bitmaps and output resolution work.
Say, I have 4k bitmaps, I make them relative to parent and set parent to 2k. What resolution will they be compressed at in sbsar?
Or if I have no external maps, and graph is relative to parent. Published file I'm using it in Painter. Will the resolution be based on painter's texture resolution?
It is not a stupid question at all 🙂
Make your bitmap node 'Absolute' and the following node in the graph 'Relative to parent'. That way when saved as sbsar the highest quality bitmap is still available when upped to 4K and all the other procedural graph elements will be recalculated based on the texture set size.
You can test the effects of different settings yourself, if you wish, with a fine grid (1 px wide black and white lines) bitmap.
Dave
Copy link to clipboard
Copied
It is not a stupid question at all 🙂
Make your bitmap node 'Absolute' and the following node in the graph 'Relative to parent'. That way when saved as sbsar the highest quality bitmap is still available when upped to 4K and all the other procedural graph elements will be recalculated based on the texture set size.
You can test the effects of different settings yourself, if you wish, with a fine grid (1 px wide black and white lines) bitmap.
Dave
Copy link to clipboard
Copied
Hey, I also have a similar question around working in scale
I have built intricate graphs in 4K, when i set the parent to 8K because i want the texutre in 8K it messes it up. I understand this is because some elements are scale dependant such as noises. However I also dont just want to set the size of the last node to parent x2 since I understood this only stretches the texture.
What are best practices to create a graph that behaves correctly when up/downscaling it between 8k 4k 2k 1k and keeps correct detailing sharp as possible? My objective is to batch export a certain texture as well as possible in the 4 resolutions. I also found having an 8K normal, moving it to a node parent/2 damages the quality of the normal, whereas recreating it keeps it much sharper.
Thank you
Copy link to clipboard
Copied
sorry i forgot to also add, how dows baking it as an SBSAR deal with this, if i change the scale of the sbsar is it recalculating the entire graph too or just downscaling in a smart way?
thanks