Getimg AI

The 38-comment USER pool (top 25 shown) splits into three clusters. HackerNews (15 threads, ~+72 each): a founder-led Show HN discussion focused on economics rather than image quality. Users calculated raw Stable Diffusion inference at ~$1/hr on an AWS g5 instance (~600 images/hr, roughly $0.0017/image) and openly questioned how a hosted service affords GPUs; the founder confirmed he 'nearly went bankrupt' offering free generations, now runs an autoscaled Tesla V100/RTX 3090 mix across runpod.io, vast.ai and Lambda Labs, and settled on subscriptions after experimenting with one-time credit packs. Sharpest criticism targets the pre-generation login wall: 'the page suggests pressing the generate-4-images button will generate four images. And then it doesn't' — one user called it a dark pattern, and a peer founder reported 60-70% user drop-off from similar walls. The disposable-email blocker also falsely flagged at least two legitimate personal domains (e.g., a three-letter .nl address), while the founder said 15% of signups were disposable multi-accounts. ProductHunt (5 comments, +1 to +4): polite launch praise for a 3-tool suite centered on an infinite-canvas AI Editor; the strongest product claim — 'best outpainting tool for Stable Diffusion right now, much better than DreamStudio' — arrives with an affiliate referral link, so it is financially incentivized. YouTube comments (5, +0): pricing backlash — one user reports a single elaborate prompt (cyberpunk skyline, '8K, ultra detailed') demanded ~1500 credits, about $7.50, 'half of the basic membership in 1 prompt', with uncertain output quality; another wrote simply 'no, not worth it.' Others want personal-LoRA/uncensored workflows ('vacation pictures I'll never get to do in real life') that no tool they tried serves. Net: a technically credible service with real friction complaints and almost no substantive image-quality feedback in the sample.

No aggregate ratings were found for this product during the last harvest.