patienceai / automatic1111-sd-2-1-banana Goto Github PK

Banana is giving me an empty response (undefined). Not an object with image(s)

Input:

{
  "endpoint": "txt2img",
  "params": {
    "prompt": "an astronaut riding a (horse:motorcycle:0.5) on the moon",
    "negative_prompt": "cartoonish, low quality",
    "steps": 25,
    "sampler_name": "Euler a",
    "cfg_scale": 7.5,
    "seed": 42,
    "batch_size": 1,
    "n_iter": 1,
    "width": 768,
    "height": 768,
    "tiling": false
    
  }
}

Is it possible to deploy this on banana but with different models that can be used via the API ? So for instance in the front-end of our app the user can select the model that will process their prompt ? eg. open journey, rev animated etc ?

Is it possible to get a list of all the accepted sampler names for this version ? eg. Euler, DPM++, Heun etc. I've tried some and I get an error, so it would be great to document what is accepted :) Thanks!

First off great repo for banana.dev it's working great!

I would like to have custom modelOutputs json returned in the return payload.

I would like all the params returned so i know what call payload was.

I forked and added this in script.py:

if mode == 'banana_compat' and 'images' in output:
output = {
"base64_output": output["images"][0],
"callParams": params
}

return output

but callParams is nill.

Any help would be appreciated. thanks -foo