phoenix
phoenixPhoenix - Your AI Assistant
Inactive session will be ended in 24 hours
End Session

Build an affordable, better-than-MidJourney image-generation bot

  • Posted:
  • Proposals: 26
  • Remote
  • #4516099
  • Open for Proposals
Emmanuel O.Anil K.Jaspreet S.Joshua A.Adnan M. + 21 others have already sent a proposal.
  • 3

Description

Experience Level: Expert
We seek a lean, affordable image-generation platform that surpasses MidJourney in UX and cost-efficiency. Deliverables: Discord/Slack bot, simple web UI for prompts and gallery, fast stylized and photoreal TTI with upscaling, variation/seed controls, prompt templates, credits-based low-cost model, NSFW moderation, community gallery, saved prompts and voting. Team: full-stack engineer, ML engineer to optimize open-source diffusion models with quantization/batching, DevOps for low-cost inference and scaling, optional prompt artist for templates and UX copy.

New Proposal

Login to your account and send a proposal now to get this project.

Log in

Clarification Board Ask a Question

  • VCONN P.

    Hi Brad, interesting project. A few questions to better understand what you mean by “better than MidJourney”:

    1. Which is the main priority for the MVP — lower generation cost, image quality, prompt accuracy, speed, or a simpler user experience?

    2. Do you already have preferred open-source image models in mind, or would you like the development team to benchmark and recommend them?

    3. What generation volume/concurrent users are you expecting initially? This will have a major impact on the GPU/inference architecture and cost.

    4. Should the first release include both Discord/Slack and the web app, or would you be open to launching one interface first and adding the others afterwards?

    5. For the credits system, do you need payment/subscription functionality in the MVP as well?

    6. Do you have a target cost per generated image or monthly infrastructure budget you're trying to achieve?

    Looking forward to hearing from you!

  • Mahmood U.

    Would you like to start with the expected monthly image volume so the infrastructure and model approach can be sized realistically?