fixed camera angle screenshot automation (for Stable Diffusion model image training)

[wikid24]

Plugin Description

last year I was able to train my own stable diffusion model on my WOL by taking 360 degree screenshots in closeup/medium/full body shots (about 40 or so in total) then creating a .txt file with same name describing each image (image 1: closeup, side view, pink hair, green eyes, elf), etc... It turned out great and was able to make a ton of SD images using my own WOL

I'm hoping to automate this somehow so I can speed up this whole process... I'm wondering if there is any interest in creating a plugin? It would help so that I don't need to spend 30 minutes to an hour manually framing the camera (poorly I might add) and taking screenshots, then having to manually create the text files.

Basically I'm looking for an addon or plugin to setup 40-50 fixed camera angles to rotate around a character a take a screenshot for me (with a fixed file name for each image). I can probably figure out the .txt file part on my own using powershell or python or something as long as the .jpg/.png filename are consistently named after the camera angle

More info

Here's some sample images (along with the contents of the .txt file) of some of the manually snapshot input training images, along with some generated images from training my own SD model:

Training Image 1:

medium shot, 1girl, bracelet, green eyes, hand on hip, hands on hips, jewelry, lips, medium hair, pink hair, skirt, solo, standing, braid, black skirt, suspenders, black gloves, sleeves rolled up, shirt, long sleeves, laces

Training Image 13:

full body shot, 1girl, braid, hand on hip, medium hair, skirt, solo, standing, thighhighs, pink hair, black skirt, suspenders, black gloves, sleeves rolled up, shirt, long sleeves, laces

Training Image 23:

medium shot, from behind, 1girl, black gloves, black legwear, black skirt, hand on hip, hands on hips, pink hair, skirt, solo, standing, thighhighs, braid, suspenders, sleeves rolled up, medium hair, shirt, long sleeves, laces

And some of the generated images I was able to create within my own SD trained model:

Original Requestor

No response

Request Terms

• I checked if this request wasn't already proposed
• I made sure this request follows Dalamud's plugin guidelines