New node: QR Code

[Sahil-Gupta584]

Closes #3468

Do you have any suggestion ?

Current Implementation:

cursorful-video-1767354987148.mp4

[timon-schelling]

!build

[timon-schelling]

I suggest adding a flag that allows merging adjacent tiles (should be on by default in my opinion).

Should be optional, because not merging can also be useful for example in this usecase.

[github-actions]

📦 Build Complete for 7073daa
https://59335d22.graphite.pages.dev

[Sahil-Gupta584]

Merged:

Unmerged:

(have renamed Merge to Merge Adjacent Tiles )

Looks good now ? @timon-schelling

[timon-schelling]

That's just merging rows, not every adjacent tile. Can you fix that?

[Sahil-Gupta584]

That's just merging rows, not every adjacent tile. Can you fix that?

1. ig think all adjacent tiles(rows and columns) are merging already.
2. you mean there shouldnt be gap btn 4 boxed-adjacent tiles?

[timon-schelling]

As you can see here in your screenshot, row are merged, collums are not merged.
I can tell because the side has indents.

[TrueDoctor]

Could we just make the shape be boxes and then use a boolean union operation to merge them? Or we could just use a manual flood fill to build contiguous regions

[Sahil-Gupta584]

I tried using flood fill approach ( ig first one will be heavier):

cursorful-video-1767593925927.mp4

[timon-schelling]

!build

[github-actions]

📦 Build Complete for de5dc82
https://0c9fbd34.graphite.pages.dev

[Keavon]

The node is currently nondeterministic. It produces different data every evaluation.

[Keavon]

See comments. The most critical one is making the node deterministic.

[TrueDoctor]

@Sahil-Gupta584 Due to the use of a Hash Set which uses randomization under the hood, the flood fill algorithm will start from a different point each time, leading to visually identical but different output in each evaluation. This should be relatively easy to fix since in some cases, the HashSet you used is just not necessary. You could for example replace the HashSet in initial calculation of the remaining pixels with a Vec etc., but please do yourself the favor of trying to understand the algorithm that is implemented and fix it without relying on ai tools for this task. The function is pretty sure and this might take a bit longer, but I think this would be a good learning opportunity

[Sahil-Gupta584]

sure, thanks for the suggestions, looks interesting for me also!

[Sahil-Gupta584]

So i tried making code more deterministic as well added few comments for understanding working.

[timon-schelling]

!build

[github-actions]

📦 Build Complete for 869a5f8
https://fa674ee4.graphite.pages.dev