This is not an AI Art Podcast: IPAdapter v2 style/composition + LoRA + Face & hand detail + Upscale
5.0
0 reviewsDescription
This is a sample workflow for the podcast This is not an AI Art Podcast (available on spotify among others, I can't link it), and the samples use the Victorian Style LoRA (available on Civitai).
These examples weren't upscaled (sorry, just whipping together examples quick!) but, they also aren't post processed. I used Juggernaut XL v9 for these generations.
This workflow is comprised of these functions:
- IPAdpater Version 2, using Style transfer + Composition transfer
- Randomized images from directory to use for many generation
- Optional to manually bypass this, see notes in workflow
- LoRA functionality
- I give an example of two LoRAs, one bypassed in the workflow. You can also bypass the one enabled one, or, use the example to add more.
- Detailer pass
- 2 passes for face detailing, disable second pass for portraits and/or change bbox_crop_factor to match.
- 1 pass for hand detailing
- SD Ultimate Upscale
- Bypassed by default, remember to enable the SD Ultimate Upscale + the save image node.
Discussion
(No comments yet)
Loading...
Resources (1)
Reviews
No reviews yet
Versions (1)
- latest (2 years ago)
Node Details
Primitive Nodes (2)
Note (2)
Custom Nodes (42)
ComfyUI
- UpscaleModelLoader (1)
- PreviewImage (8)
- CLIPTextEncode (3)
- ConditioningCombine (1)
- EmptyLatentImage (1)
- CheckpointLoaderSimple (1)
- LoraLoader (2)
- LoadImage (2)
- VAEDecode (1)
- SaveImage (3)
- KSampler (1)
- VAELoader (1)
- SAMLoader (2)
- UltralyticsDetectorProvider (2)
- FaceDetailerPipe (1)
- FaceDetailer (2)
- FL_ImageRandomizer (2)
- IPAdapterUnifiedLoader (1)
- IPAdapterStyleComposition (1)
- PrepImageForClipVision (2)
- Seed Generator (2)
- UltimateSDUpscale (1)
- Image Filter Adjustments (1)
Model Details
Checkpoints (1)
juggernautXL_v9Rundiffusionphoto2.safetensors
LoRAs (2)
gildedvictoriansxl_v2-000002.safetensors
guideboatxl_v1.safetensors