Optimal parameters
Hi, I had a few questions about the ideal parameters for QR generation in order to work best with this model.
I'm assuming the module size of 16px is relative to a total resolution of 512x512 right? You would want to optimize the error correction to maintain those block sizes?
Is there a specific border size that is optimized for the model or has it been trained on many border sizes?
What do you mean by "relative to a total resolution of 512x512" ? "Module size of 16px" is just that, 16px per module, whatever the resolution. You can try other sizes but it was optimized for 16px.
For the error correction, try to set it as high as possible, but sometimes lower is better when it makes your code bigger.
The borders should not be a problem, whatever the size. I would say it's better with a multiple of 16px, but it should work either way.
Oh wow. I did not realize your model was that flexible! I just assumed it had to be in the native resolution of the output. I was frustrated at confining myself to lower error corrections to fit extra self imposed constraints. Excited to see what can be done with proper use! Thank you!
Oh actually I meant "whatever the resolution of the output", but you should always match the controlnet and image resolutions with this model.