It's not a secret anymore...
Take your Secrets game to the next level using one of SuperSecrets' powerful tools
SuperSecrets helps you create and refine high quality image prompts using Secrets-specific research, privacy-conscious AI processing, and practical feedback you can learn from and use straight away.
Privacy first
Prompts run through Venice.ai, helping keep your ideas private and away from third-party provider training or harvesting.
Secrets-focused
Tuned through months of testing, iteration and research specifically for the Secrets.ai image platform. No other platform has this level of knowledge.
Free with BYOK
Bring your own Venice key for unlimited use of the tool!
Subject to your own Venice inference limits.
Compose a new prompt
Build structured, polished image prompts from scratch, with stunning results
Improve an existing prompt
See your prompts' strengths and weaknesses, with practical advice for more effective prompting.
Design nuanced characters
Use innovative tools to create deep, nuanced characters that captivate the imagination.
Discover helpful tools
Need a quick bad word checker or character counter? Howabout a focused image describer?
More tools — and an even better homepage — coming soon.
Changelog:
- Some cosmetic refinements of the homepage (still not finished or anything, just got sick of seeing the old one)
- Made light mode a little bit less intense by darkening the background of non-interactive areas.
- More site-wide cosmetic changes/improvement. Really padding out this changelog!
- Some internal code cleanup and routing changes to prepare for Character Designer. Hopefully nothing is broken! 🤞
- Made an adjustment to (hopefully) further help explicit images be a bit less error-prone (logistically-speaking)
- Slight adjustment to tone down heavy makeup being applied frequently by default
- Least exciting update ever ... just fixed a couple of minor bugs reported by users.
- Toned down "Calm" flavour's desire to desaturate images
- Forced a check & removal for HTML tags in the final prompts
- Updated flagged words
- Added ability to revert back to the previous iteration when building a prompt and making changes via iteration.
- Fixed some text and logic issues.
- Further tweaks to flavours to achieve a bit more balance between the options.
- Added ability for users to set their own default flavour option under the user settings section.
- Added a new 'Calm' flavour for more grounded imagery.
- Fixed a bug around additional instructions not always being passed through.
- Vastly improved reference images, added 'focus' functionality so that the analysis can be tailored to the specific area of interest within the reference image.
- Added standalone image analysis tool
- Added notification system to warn of upcoming downtime etc.
- Added X-1 experimental flavour, currently using Venice's Uncensored Roleplay model
- Added a 'prompt title' and 'opinion' to the final stage of image prompt building
- Styling, layout and wording tweaks
- Make image analysis place less importance on things that will largely be out-of-frame in some situations.
- Minor bug fixing and wording tweaks
- Fix minor bug introduced by bad-word checking.
- Made the system be more inclined to include footwear when building outfits.
- Added standalone Bad Word Checker tool. Re-jigged the menu items and redesigned the menu to have a better layout on mobile.
- Added Bad Word checking to final output. Bad words are usually filtered out but sometimes slip through, this will highlight them.
- Made prompt summary a bit more obviously part of the prompt breakdown.
- Some other minor bugfixes
- Added an attempt at better handling for explicit images.
- Fixed a bug when logging in without a valid API key being present.
- Added ability to reset prompt session.
- Cleaned up some dark mode bits.
- Squashed a minor session bug.
- Yay! What could possibly go wrong?