Fast Creator v1.0.4 (Update 21/06/2024)
5.0
1 reviewsDescription
Fast Creator
New Version Called
History
This Workflow was born out of the need to have batch images of a minimum of 8 to a maximum of 20, with sizes from 256x256 to a maximum of 512x512 pixels, but the creation had to be fast, so I had opted for SD Hyper, modified to a turbo one for graphical reasons.
I had done it to have some basic images to pass to another workflow that did upscaling and added some detail. After a while of using it, I took the step of unifying the two workflows, to have a "convenient" generation, and hence version 1.0.2.
But I wanted a little more detail, especially when I was creating batches of 256x256, so I upgraded this workflow once again to v1.0.3 (current version).
Workflow Details
Workflow is very fast in generating draft images at sizes 256x256 and 512x512; the number of batch generations is adjustable, as are the sizes.
One thing I wanted from the beginning, was to decide as well the size of the Output file, so whether the batch was 256 or 512, or any other size, in the output always had to remain a generation of the desired size.
In this version I have also added the choice of being able to input an image in order to add details to an existing image, without the need to create new ones.
How does it work?
After generating the images for the batch, they are displayed in a node act to choose the one that is most to our liking.
After choosing the best image and clicking on it, the workflow processes it, passing it through various stages, including 2 Ksamplers for details, 2 upscaler netodes for enlargements, some Face detailer and hand detailer, at the end you have the intermediate images of the process, and the final images, with comparator.
final conclusions on Workflow
As can be seen from the images, the details are quite good if we compare the draft image with the finished image.
A few more images












Disclaimer
All images reproduced here are images created directly from my prompts and with this workflow.
All prompts or images on this space, are original, created specifically for this, and on this, workflow.
Updates and Help
In the next few days I will see about putting a guide directly within the workflow, so that help is always at hand.
Change log
21/06/2024 Version 1.0.4
I'm starting to enter the comments for the guide
Added the ability to choose whether you want to prioritize Graphics or Speed.
Choosing to prioritize Graphics will result in images with higher quality and more detail, at the expense of process speed.
This is very useful if you create batches of 256x256, the images you will get, will be more detailed and high quality than the lesser quality and detail of the Speed mode.
Deciding to prioritize speed will result in images with lower quality and less detail, but will result in faster processing speed.
If batches of 512x512 images are used, the quality and detail that will be obtained with the Speed mode will still be greater than the quality of an image resulting from a batch with 256x256 images.
______________________________________________
Below are some examples of comparison between the two modes
the image above (right image), was obtained from an image of size 256x256 (left image) in Speed mode
above the same image as before (256x256 , left image), but in Graphics mode (right image)
Below are the finished images.
Speed Mode

Graphic Mode

Discussion
(No comments yet)
Loading...
Reviews
No reviews yet
Versions (2)
- latest (2 years ago)
- v20240618-231347
Node Details
Primitive Nodes (45)
DF_Integer (3)
Display Int (rgthree) (2)
Fast Bypasser (rgthree) (1)
Fast Groups Bypasser (rgthree) (1)
GetNode (4)
JWFloat (1)
JWInteger (1)
KSampler Config (rgthree) (1)
Label (rgthree) (5)
Note (9)
Power Lora Loader (rgthree) (3)
PrimitiveNode (4)
Reroute (6)
SetNode (3)
Switch image [Crystools] (1)
Custom Nodes (182)
- ImageNoiseGaussian (1)
- ImageEffectsAdjustment (2)
- CR Text Concatenate (1)
- CR Simple Image Compare (2)
ComfyUI
- ImageUpscaleWithModel (1)
- UpscaleModelLoader (1)
- CheckpointLoaderSimple (1)
- VAEDecodeTiled (2)
- CLIPTextEncode (7)
- KSampler (3)
- PreviewImage (14)
- VAEEncodeTiled (2)
- ImageSharpen (2)
- VAEDecode (1)
- ImageBlur (1)
- LoadImage (1)
- EmptyImage (1)
- EmptyLatentImage (1)
- CLIPSetLastLayer (1)
- SimpleMath+ (2)
- ImageResize+ (1)
- GetImageSize+ (1)
- UltralyticsDetectorProvider (3)
- FaceDetailer (3)
- ImpactMakeImageBatch (1)
- ImageScaleDownBy (1)
- Big Image Switch [Dream] (3)
Hakkun-ComfyUI-nodes
- Any Converter (2)
- Preview Chooser (1)
- ColorMatch (1)
- mape Variable (104)
- FaceFixerOpenCV (2)
ntdviet/comfyui-ext
- gcLatentTunnel (3)
- MathExpression|pysssss (1)
- SDXLPromptStyler (1)
- SDXLPromptStylerbyLighting (1)
- Seed Everywhere (1)
- JWImageResizeByLongerSide (2)
- SimpleText (1)
- Text Multiline (2)
Model Details
Checkpoints (1)
SDXL\dreamshaperXL_v21TurboDPMSDE.safetensors
LoRAs (0)