HiDream - Txt2Img with Hi Res Fix & Upscaler.
5.0
0 reviewsDescription
This does work with all 3 models, although Dev does seem to work best with this.
For HiDream Full use uni_pc sampler, simple scheduler, 50 steps on the first pass, 20 steps on the second pass and upscaler.
For HiDream Dev use lcm for the sampler, normal for the scheduler, 28 steps on the first pass, 8 steps on the second pass and upscaler.
For HiDream Dev use lcm for the sampler, normal for the scheduler, 16 steps on the first pass, 8 steps on the second pass and upscaler. (I found going below 8 on the 2nd and Uspcaler stages to drop the quality).
There are 3 bookmarks set:
1 = Prompt Groups
2 = Image generation & settings
3 = Upscaler
There are 4 prompt groups to choose from, use the Fast group bypasser to select the one you want and the switch to the right of that to select the correct group.
1 - Manual Prompt (Default)
2 - Portrait Master & Magic Photo Prompter
3 - LLM Prompter (Requires Ollama)
4 - Vision LLM Prompter (Requires Ollama)
There is also a TTP Upscaler. It works, but can be a little hit and miss, most of the time I have this group switched off.
Discussion
(No comments yet)
Loading...
Reviews
No reviews yet
Versions (1)
- latest (8 months ago)
Node Details
Primitive Nodes (32)
Bookmark (rgthree) (3)
DepthAnythingV2Preprocessor (1)
DetailDaemonSamplerNode (1)
Fast Groups Bypasser (rgthree) (2)
GetNode (2)
Image Comparer (rgthree) (2)
LayerFilter: FilmV2 (1)
Magic Photo Prompter 🪄 (1)
ModelSamplingSD3 (1)
Note (3)
OllamaConnectivityV2 (2)
OllamaGenerateV2 (2)
OllamaOptionsV2 (2)
OverrideCLIPDevice (1)
Power Lora Loader (rgthree) (1)
Primitive float [Crystools] (1)
QuadrupleCLIPLoader (1)
SetNode (1)
Starnodes_Aspect_Ratio_Advanced (1)
TTP_Image_Assy (1)
TTP_Image_Tile_Batch (1)
TTP_Tile_image_size (1)
Custom Nodes (48)
- CR Text Replace (3)
- CR Text Input Switch (4 way) (1)
ComfyUI
- LoadImage (1)
- SamplerCustomAdvanced (3)
- LatentUpscale (1)
- CLIPTextEncode (2)
- VAEDecode (2)
- VAEEncode (1)
- BasicScheduler (3)
- PreviewImage (5)
- UpscaleModelLoader (1)
- ImageUpscaleWithModel (1)
- VAEDecodeTiled (1)
- ImageScaleToTotalPixels (1)
- SaveImage (3)
- RandomNoise (1)
- CFGGuider (1)
- KSamplerSelect (1)
- VAELoader (1)
- UNETLoader (1)
- easy portraitMaster (1)
- SimpleMath+ (2)
- ImageListToImageBatch (2)
- ConcatText_Zho (2)
- ColorMatch (1)
- ShowText|pysssss (3)
- String Literal (3)
Model Details
Checkpoints (0)
LoRAs (0)