Sign In

Flux.1[dev] Img2Img + GPT-Interrogator + 2K Upscaler Workflow

47
827
20
Updated: Aug 23, 2024
toolworkflowflux.1flux1.d
Type
Workflows
Stats
827
0
Reviews
Published
Aug 5, 2024
Base Model
Flux.1 D
Hash
AutoV2
92A55C67D7
default creator card background decoration
Binx's Avatar
Binx
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Auto-installing the nodes doesn't install my fork, which fixes issues with the original GPT prompt node!

This workflow uses Flux + Img2Img + GPT-4o-mini to create and upsample provided prompts. It works incredibly well in my experience.

Optional nodes in top-left for clip merging old loras for experimentation.

The GPT-Interrogator can be found here

The GPT-Interrogator was a modification of the ComfyUI-KepOpenAI node pack, which I've only just made into a fork because the original repo isnt maintained. The only changes made were to nodes.py. Prompt below:

I am going to give you an image as well as some keywords. use both of them to create an elaborate generative image prompt using approximately 256 tokens, with exquisite detail. Use complex and uncommon descriptive words. Favor heavily stylized image descriptors. If needed, weights can be applied to the caption in the following format: '(word or phrase:weight)', where the weight should be a float less than 1.8. Only reply with your proposed text prompt and nothing else. Keywords: <insert keywords here>