try new Seedance video model that can combine image elements with character consistency #photoai

tell FAL I want a discount now that I'm on the new GPU platform succesfully so I have more negotiation power #photoai

fix ImageMagick sharpen out of memory error for giant pics #photoai

fix downgrade plan didn't work #photoai

try Wan-2.2 Fun Control model #photoai

change model training steps back to how it was before churn started to test #photoai

publish Claude artefacts to rank on Google (secret new trick) #photoai

reach 86% profit margins #photoai by switching to new API provider and saving $20,000/mo in GPU costs

get Flux training working on new cheaper platform thanks @cheepo2109 #photoai

move Flux Lora Fill to new cheaper GPU provider #photoai

save $3K/mo on Flux training moving to new GPU platform that's 66% cheaper per training #photoai

switch to new GPU platform and get 25% cheaper LoRa image generations and save $4K/mo on $13K/mo in image gen bill #photoai

deploy Nano Banana new edit model by Google on #photoai

try Nano Banana with input pics of person and then ask it to take a new photo to see resemblance and it's not high resemblance but okay #photoai

try Nano Banana #photoai

try Nano Banana new mysterious image model #photoai

try first and last frame same again #photoai

run new talking video workflow #photoai

change talking video pipeline for [ Talking video ] on static image to not use Omnihuman (img2video) but instead first make a video (img2video) then Lipsync it #photoai

try video model with same first frame + last frame so we can make better looping talking videos #photoai