Multiple angle photo shoot creator
This workflow uses a detailed initial prompt to generate an initial image, then uses "ReferenceLatent" nodes and a consistent seed to generate variations with a consistent character.
The workflow allows you to direct how each image is modified to create the pose/effect you're after in each individual picture. Type the modification prompts as if you're telling it to do things, I've included the prompts for the exemplars in the workflow.
The upscaler uses a similar philosophy to modify lighting/look/feel to be consistent across the "photo shoot" set.
Models used: SNOFS for the initial image generation, then Flux2 Klein for the upscaling.
The workflow is designed to cater for multiple GPUs (as that's what I'm using), if you only have one then either set the number of GPUs to 1 or replace the nodes with the single GPU version.
I have included a place in the workflow for people to drop a LoRa in, though they are bypassed by default.
Edit: I'm working on a v2 of this with 4k upscaling and more steps to modify generated images. This allows one to perfect a 4 photo set with prompts alone (no inpainting etc.).







