Use seed and prompt in image filenames about invokeai HOT 9 CLOSED

I'll be adding this feature in next 1-2 days, right after adding k_lms sampling

from invokeai.

I'm creating a new issue for this. I've actually never tried to save the images form Discord. Do you right click on them or something, and is the filename automatically assigned when they download?

And you want this for the vanilla txt2img, since that's working for you?

from invokeai.

Yes filenames are like this https://cdn.discordapp.com/attachments/1005596318267363488/1010069669182459934/macro_portrait_of_werewolf_in_times_square_painting_by_frank_frazetta_by_artgerm_-C_8.0_-n_4_-g_-S_1287913343_ts-1660889775_idx-0.png

Im gonna Use Your bot code cause the vanilla code is too limited ot use conveniently, so its up to You really if You want to dabble with 2 branches, i dont see the advantage of having to wait for model to load for a couple minutes everytime you want to run prompts vs loading it only once.

from invokeai.

Thanks for the example. That's a freakin' long file name! How about if I write the prompt into the file itself? You can put arbitrary text into PNG files and either pull them out with a script or view them in a PNG GUI.

from invokeai.

thats even better, not a fan of long filenames

from invokeai.

Lot of people assume the model will drop tomorrow on github, we will see...

from invokeai.

I love the idea of writing the prompt (along with the seed) as meta data in the png file and creating file names that are not so unweildy.

from invokeai.

The prompt and all its switches are written into the PNG metadata as a tEXt string labeled "Dream" as of version 1.02.

from invokeai.

from invokeai.