Textual Generation Data Import (like provided in CivitAI and stored in images) for Generate Tab

[onliner2021]

Feature Idea

I am missing an important feature in SwarmUI (or maybe I just haven't found it yet :-). In A1111/Forge/SDNext etc. I was always able to import a text prompt from a generated image. Not just the positive/negative prompt, but applying all parameters (width, model, upscaler, clip skip,...).

These infos are stored inside the PNGs/JPGs in most generations. And these textual infos are also available for posted images in CivitAI on the right side:

In A1111 it is possible via:

In SDNext via the restore button:

For example, such a text with prompt data and parameters: The Northern Lights illuminate the night sky<lora:Anime_art:1.2> Steps: 30, CFG scale: 3, Sampler: DPM++ 2S a_ddim_uniform, Seed: 124263196777517, VAE: ae.safetensors, Size: 1024x1024, Model: flux1_v30Fp8.safetensors, Model hash: b0b64b4b6a, Hires upscale: 0.4000000000000001, Hires upscaler: remacri_original.pt, Denoising strength: 1.0, Clip skip: 1

Is there a way to do this in SwarmUI? I know, that I could import JSON or CSV files via Import Presets. But that's not what I am looking for (as JSON or CSV is not the native format in which CivitAI and all other AI Generation tools persist this data by default. I also know that I can drag an image into the Comfy Workflow windows and it would read out the info from the commend or EXIF data. But some images just don't have that and all one has is a text (from the clipboard). Also importing it into the Comfy Workflow is not what I am looking for, as the generation parameters look totally different than the normal "Generate" parameters.

I am hoping that there is a way to paste/import such a text info somewhere in the normal Generation tab of SwarmUI and have it interpret that data. And auto-fill the negative, positive prompts. As well as adding the LORAs that are referenced in prompts e.g. <lora:Anime_art:1.2>. And setting the parameters to the given values.

That's the only missing thing right now, that is keeping me from switching to SwarmUI all together at the moment. I think SwarmUI is superior to all of the other SD/FLUX UIs out there in many features (except ComfyUI, but since SwarmUI has integrated ComfyUI as well it's the best of both worlds). But it's important to be able to remix existing images from CivitAI with it (without having to set all options manually yet again).

It would of course be ideal it one could just import these textual settings via the "Import Preset" feature (so that it would support JSON, CSV and also the "standard" textual representation.

Thanks for considering (and maybe it's already there, and I just haven't found it yet).

Other

No response