Open Carlmcap opened 2 years ago
I had the exact issue, and this solution below solved it (and it seemed to have solved the other users' issue as well):
https://github.com/HashLips/generative-art-opensource/issues/49#issuecomment-913864196
@Devolvera Thanks! I am trying this now and it works, ill run some analytics when it finishes to determine relative rarity from this method. Ill report back.
Also, Are you using V3, or main? which is better?
@Carlmcap Awesome! I'm glad it works, and yes, I would love to see the rarity distribution as well.
I used V4 for this project
@Devolvera
Interesting stuff. It looks like the following with 727 runs:
Layer 1: 12 Options -- H (weight 2) 15%, L( weight 9) 7.1% Layer 2: 6 Options -- H (weight 1) 26%, L (weight 3) 13.7% Layer 3: 3 Options -- H (weight 1) 52%, L (weight 3) 22.5% Layer 4: 5 Options -- H (weight 1) 32%, L (weight 5) 15.6%
I may try to tackle the main one with the % toggles tomorrow....
4 had the highest rarity, but nothing above BASE layer 2 weight 1.
So in this instance. Layer 2 is the base of the photo. 1 being common, 6 being rare.
Only Layer two weight 1 contains the perfect score (4 different times) .
Lastly, my code infinite looped at 727 PNGs.
Hello I am running image generation with 4 layers.
I am receiving the DNA Exists!! Log after 25 runs. I cant seem to get more running.
Also, It seems the weight is not working properly for me. a majority is weighted 100, with some select items weighted 5-50. But I get a lot of the low weights in the 25 output prior to loop.
Let me know your thoughts!