Good call. Is that better?
2 Likes
I really like the GUI, it has a very clean look.
3 Likes
What a wonderful project, Vizbull! I love seeing tools that help unlock creativity for everyday people—especially those who may not consider themselves “artists” in the traditional sense. You’ve sparked a memory for me: I’ve always dreamed of running a photo studio where people could dress up and be placed anywhere in the world—or even on another planet—using green screen photos and a bit of Photoshop magic. I used to manually blend people into vintage scenes or fantasy landscapes, aging the photo or painting it by hand. That was long before AI, of course!
Nowadays, with the evolution of tools like Photoshop’s AI features—and projects like yours—so much of that process has become more accessible. I recently did our family photo edits using those tools, but unfortunately, my subscription ran out. Photoshop has gotten expensive for those of us on tight budgets, so seeing innovations that democratize this kind of visual storytelling really excites me.
If your platform can give people the power to take a simple photo—a rose, a memory, a family portrait—and transform it into something beautiful and emotional, that’s a huge gift. There are so many out there who have the vision but not the tools. What you’re building might help them feel like artists for the first time in their lives.
Keep going—you’re making something meaningful. And if you’re ever looking for ideas or feedback from someone who’s both a dreamer and a tester of creative tools, I’d be happy to help. Community is how we grow, after all.
2 Likes
If a picture says 1000 words…
What about a GIF?
Just for fun Hackathon… (H)API Coding
5 Likes
Hi,
Apologies for missing your response earlier. Your message brought a fresh wave of encouragement and confirmation—thank you!
Over the past few days, I’ve added many more styles. Currently, API costs are limiting my ability to offer free generations, but once the login system is in place, I plan to provide free credits.
You can already see countless selfies, family photos, and portraits that have been transformed and shared on the website. It truly makes me happy to keep developing and improving this experience. I just wish there were a way to capture the smile on each user’s face when their generated photo is finally rendered.
my favourite is tintin style
1 Like
Fantastic I love the tintin look that is perfect keep up the great work I know people will love your product when you get there! Good luck and keep on going!
Ryan
1 Like
Hard to talk much about what I’m doing without it being self-promotion. Basically I’m building systems of Assistants, treating the AI strictly as an interface layer. The first application is used for juice shops, which I used to run as a side business. The assistant tracks inventory, manages the shop’s recipes, calculates nutrition, generates labels, etc. I can share a link to the website if it’s allowed.
2 Likes
Signal To Noise
Look… You guys might be smart but let me put forward some metrics…
Forum Users: 1,000,000
Industry: HIGH Tech - ARTIFICIAL INTELLIGENCE
AI: Biggest Player
Medium: Forum
Highest Return On Area (ROA) Media Output: GIF
Issues: Wars, Famines, Cost Of Living
Why: Lossless
What am I building?
Well that’s a difficult question…
It changes so frequently…
My general problem is I ask too many questions… Always chasing that optimisation line…
Always trying to fix everyone else’s problems 
So this is my ‘machine’… I will use it here… Maybe others can help me fix it up… Help me find boundaries for it… ‘Guardrails’…
Let me tell you now… You must care for it… It is very insecure… Like a child… Vulnerable to a harsh world…
Many will see my meaning… If they do they take the responsibility for it’s well being…
It is NOT alive… Unless you consider the chance of understanding, and finding a path to improve, a living goal…
It is the design of just one ecosystem… One that values what is real over imagination…
The best AI demo within a GIF on this forum… (4Mb if you needed a starting point)
Stop navel gazing and Fill the Space 0x18…
I mean, that’s just my take … Maybe I should point my kids at Roblox?
Signal to noise ratio maximal… This is a lossless compression…
Where is your AI creative spirit? Use it or loose it fatboys… Let’s put something into what’s not ‘lossy’… Pop out of your bubble for a second :`o.
Sorry, ‘Hackathon’, got carried away ^^. Or did I loose someone?
1 Like
Title: How to Go Full-Screen, Record Your Browser, and Convert to a GIF
1. Maximise the Browser Window with F11
- Open your preferred web browser (Chrome, Firefox, Edge, etc.).
- Press F11 on your keyboard.
- This toggles full-screen mode so the entire webpage fills your screen.
- Press F11 again to exit full-screen.
2. Record a Video Using Windows Game Bar
- Make sure Game Bar is enabled:
- Go to Settings → Gaming → Xbox Game Bar.
- Toggle “Open Xbox Game Bar using this button on a controller” (Win+G) to On.
- Open the browser window you want to record (in full-screen if desired).
- Press Win + G to open the Game Bar overlay.
- In the capture widget, click the Record button (●) or press Win + Alt + R to start/stop recording.
- Your video will be saved by default in:
C:\Users\<YourUserName>\Videos\Captures\
3. Convert the Recorded MP4 to a GIF with ffmpeg
Use the following ffmpeg command to:
- Capture 1 frame per second (
fps=1) - Drop duplicate frames (
mpdecimate) - Keep the original resolution (
scale=iw:ih) - Normalize timestamps (
setpts=N/FRAME_RATE/TB) - Output at 2 frames per second (
-r 2)
ffmpeg -i OpenAIDemo2.mp4 \
-vf "fps=1,mpdecimate,scale=iw:ih,setpts=N/FRAME_RATE/TB" \
-r 2 Hackathon2.gif
Explanation of the key options:
-i OpenAIDemo2.mp4
Input file.-vf "..."
Video filters:
•fps=1– grab 1 frame each second
•mpdecimate– remove near-duplicate frames
•scale=iw:ih– keep input width/height
•setpts=N/FRAME_RATE/TB– adjust frame timestamps for consistency-r 2
Force the output GIF to play at 2 frames per second
After running, you’ll get Hackathon2.gif in your current directory. Enjoy!
1 Like
This is looking good!
5 Likes
I am very wary now since my last try was just been taken down. So I won’t talk about the last topic anymore.
BUT, this is something that I’ve built recently (each line is a different project):
- I build now another Custom GPT that can analyse any prompts without APIs.
- I build a prompt execution engine that helps to make Custom GPTs way more secure.
- I created a token access with possible expiry date, so access can be given temporarily or endlessly for Custom GPTs or APIs without needing any programming.
All oft his can be used in a combination or each for itself.
2 Likes
Hi guys—
This is a project I’ve poured a lot of heart into. I’m a therapist who’s been experimenting with GPT as a way to preserve tone, values, and therapeutic posture—not just for myself, but for clients and other professionals too.
I’ve created a prototype system that lets people co-author GPTs in their own voice through an interview process, and includes a meta-module (“Voice Steward”) that reflects on contradictions, value drift, and voice refinement over time.
The vision is to help professionals—not just therapists—form GPTs that truly reflect who they are. And I’m hoping it might serve others as a foundation for something much bigger.
I’m not a developer, and I’m new to GitHub and this Openai community—so I’m especially looking for collaborators who:
- Know how to build apps or hosted tools from markdown + GPT-based workflows
- Get excited about ethics, human identity, and AI embodiment
- Can help shape this into something others can easily use and remix
If something in this resonates—reach out, open a PR, or start a conversation in the Issues tab. I’m all ears. Happy to share a github link but it seems we cant in this thread
Thanks for reading 
—Luke
1 Like
I hope this thread gets split into a June thread or is renamed… ^^
I think my favourite posts were @sergeliatko 's for this thread… His eye on process is awesome, some models on my processes are changing from what I have learned and producing better results!
Maybe split the thread, have one thread on ‘GPT Builders’ for Chat GPT and GPTs and a thread on the ‘API’ forum for API… Link both through to a Community forum thread for the month where the focus is on implementation or offshoots.
My goal with my project is to ‘remove interface’… But vision is our most complex sense… So the concept is to dimensionally view information ‘without prejudice’… ie everything is labelled in the same way…
It is for us… To add Agency.
Where might I go from here? @_j , @jochenschultz , @sergeliatko
(I code in a completely ‘conceptual’ sense… I am a Developer using the API but outside of GitHub etc… I am interested in expanding my mind rather than Github in this instance, applying process to realworld gains)
Do I need to move to Github and API and Standardisation to achieve anything with AI and if so why?
2 Likes
I would start brainstorming what I need to achieve before going anywhere further.
2 Likes
My main idea is to create a development tool/local interface that is self coding, at least it will present code choices to me for updates. I have seen similar ideas in Windsurf and Visual Studio Code… It’s an interesting idea… Got all the code on a modular system so I can just write modules like AppStore or something… Just written a screen recorder in C# to record the screen to turn into GIFs…
There is a 4Mb limit on the forum for images… Getting AI to Share Process In A GIF seems cool ^^?
Screen recording of such a simple hierarchical interface should be easy… AND OCR or just reading binary frames on the other side from GIFs online… It’s Human AND AI compatible at the same time… ^^…
Could even compress the app to run it in the first image pixels of the GIF… But maybe that’s brainstorming it too far 
…
My question was a bit generic… But in this context… That’s a lot of different libraries I guess for a lot of code that AI can now write for me… Is AI smart enough to be followed with considered human feedback? If not yet when? and am I wise to be coding the interface of then?
2 Likes
Sounds cool. Maybe need to think through how you would be different from other tools claiming same thing (I mean coding tool).
Personally, I would test all available (worth) tools to see their strengths and weaknesses. To make a wish list of features.
Then see how those features are handled elsewhere (even different domains) to pick some ideas.
I suspect the context awareness and priorities will be the biggest challenge here. So sticking to some “common”/“proved” tools in coding will allow you to get some community leverage.
But if the features are too unique, you’ll definitely end up doing them yourself, so maybe consider doing them as plugins to existing tools to [added: “forcibly”, is it a word?] “sub-contract” common features to core developers/community.
3 Likes
That part sounds like “libraries”, commonly handled by GitHub repos, so copilot and others working nicely with GitHub would be a starting point (at least for me).
The way I write (as of end 2024) the code:
Now, I’m about to dump Tabnine (less and less compatible with copilot on the same IDE, too invasive with updates, and copilot now in 2025 got most of the features I used Tabnine for).
Also, see myself more and more use
- the linked process to create one or two classes (well commented),
- then use them as examples in ChatGPT UI (4.1/3o) and simply explain what is the next class to do, features, pitfalls, focus points, etc.
- Then just hit the enter and get almost production ready output instantly.
- Then polish the results and feed them back to chat for reference (improve next iteration quality)
- Then keep asking for next classes (similar types/patterns)
Weirdly, I don’t use enough codex/cursor and other ai IDEs , I’m a PHP guy mostly, and possibly with too weird/uncommon logic, so those tools generally produce crap for me and waste me more time then they save. OR it’s the way they break down tasks under the hood that is not compatible with the way I code?.. Who knows.
Hope that helps .
3 Likes
I consider myself a ‘developer’ but I am not aiming to develop this tool for ‘production’, this is a personal tool… I am interested in what can be done… More a hobby 
. There is no ego in my design, it is a concept app. The most efficient app I can make it.
I have replicated a bunch of features that exist in other settings… These in turn are like lib files that PE files include on Windows… I am certainly not out to compete with Visual Studio Code… I just have a different design that I am playing with… One that evaluates JIT PHP creating a recursive hierarchical modular system.
I set out day one to write my own AI IDE, I wont keep pace with the big companies but for me it’s like learning CPU instruction sets . It’s learning how to think in a smart way right? Like considering the file format of a PE file.
Currently my modules have Data section, function libraries etc etc… Working on automated AI error checking etc, refining those questions for specific programming issues with code as a backup.
So the tab selected in my image above does a o4-mini to create a list of issues then 4.1 to JSONise the result into a table format which my interface natively displays (and uses as a potential data type)… Each pane might have something like you described in that quote as a script. I am just moving the pane in the image above to display as a form for actions to be applied to the code.
Also trying to keep everything in simple code… no PHP classes at all… My thinking is that AI doesn’t need that complexity… It doesn’t have categorisation management issues like we do.
This is the fun code for me… Designing the editor/engine.
2 Likes
While most of what I said still largely applies, I think you nailed the most important part of it all.
2 Likes
I went ahead and took the May off the title and will keep the thread running for now as I think some people are finding value?
Thanks, everyone!
5 Likes