Changed Surface blur diameter to radius #1978
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
I'm fine with it just breaking a breaking change tbh |
Then let's do that. |
joeyballentine
approved these changes
Jul 24, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I noticed that the Diameter input of the Surface Blur node (#1292) behaved weirdly. In my testing, I set both sigmas to 1000 (max), which makes the node behave like a simple Gaussian blur. This makes it easy to see the effects of the diameter. I saw that D=1,2,3 produced the same image. Same with D=4,5 and D=6,7 etc. In general, only odd numbers >=3 were actually used by OpenCV's
bilateralFilter.To fix this, I switched to radius instead. The diameter is then calculated as 2r+1. This makes the node behave as expected and is consistent with all other blur nodes. I also set 100 as the max radius, because Surface blur is extremely slow for large radii (anything R>30 takes multiple seconds at 100% CPU on all cores).
Since both linear and log scale didn't feel right for the radius slider, I added a new sqrt scale. We can use this scale for other stuff in the future, but it's only used here right now.
Migration: Since this change is breaking, I added a migration. Migrating input values is easy, but connections are hard.
In fact, 3 nodes necessary to approximately convert from diameter to radius (D=1 is incorrectly mapped to R=0):
And it takes 6 nodes for a correct conversion:
Do I have to code up a migration for connections, or can we just list this under "breaking changes" in the next release?