QWE AI Academy You made the caricature. The one everyone’s posting. ChatGPT turned your selfie into a cartoon version of you surrounded by laptops, coffee mugs, whatever screams “this is my job.” You posted it. Got likes. But then you thought: what if I pushed this further?
That’s the question most tutorials skip.
The Iterative Trick
I generated my caricature using the standard prompt. Then I typed: “Make it more ridiculous.”
ChatGPT didn’t ask questions. It just went. The laptop on my desk doubled. The coffee mug became a tower of mugs. My exaggerated features got more exaggerated. I did it again. “More ridiculous.” Now I’m drowning in a sea of code snippets and the coffee tower is on fire.
This is ChatGPT’s image editing workflow applied to viral content. You’re not starting over – you’re iterating on the previous output. Each time you say “make it more ridiculous,” ChatGPT treats the last image as the base and escalates one step further.
But is this just parlor tricks, or does iterative escalation actually match how creative work happens? Comics iterate punchlines. Designers push mockups. Turns out, AI can do the same thing – if you structure the conversation right.
How to Actually Do This
Step 1: Generate your base caricature. Upload a clear, front-facing photo to ChatGPT (as of February 2026, this viral trend uses the prompt “Create a caricature of me and my job based on everything you know about me” – it pulls from your chat history). If ChatGPT doesn’t know you, specify your job and a few personality traits.
Step 2: Stay in the thread. Most people download and stop. Don’t.
Step 3: Type “Make it more ridiculous.” That’s it. No need to re-upload or clarify. ChatGPT will grab the image it just made and amplify the absurdity.
Step 4: Iterate. Keep going. “Even more ridiculous.” “Turn the absurdity up to 11.” “Make it completely unhinged.” Each iteration builds on the last.
Pro tip: Want to steer the chaos? Specify what to exaggerate. “Make the background more ridiculous” or “Add more objects related to my job” gives ChatGPT direction without killing the momentum.
What Happens After 3-4 Rounds
Around the third or fourth iteration, my face started drifting. The exaggerated cartoon version of me from round one? Unrecognizable now.
Why? ChatGPT’s image editor doesn’t keep referring back to your original uploaded photo. (OpenAI’s docs confirm this – it generates new images based on the previous output and your new prompt.) After several rounds, you’re iterating on an iteration of an iteration. The original reference is gone.
Two options: stop after 2-3 rounds, or re-upload your original photo mid-workflow and say “keep this face but make the scene more ridiculous.”
The Daily Limit
I hit this on round five. Typed “make it more ridiculous,” hit enter, and got: “You’ve reached your image generation limit for today.”
No warning. No countdown. Just a hard stop.
Free ChatGPT users have a daily cap on image generations (OpenAI doesn’t publish the exact number, but community reports as of early 2026 suggest 10-15 images per day). ChatGPT Plus ($20/month) gets you more, but there’s still a fair-use limit. If you’re iterating, you burn through your quota fast.
The workaround? Spread iterations across days, or upgrade to Plus if you’re planning to go deep.
Why This Technique Actually Works
Most AI image workflows are one-and-done. You write a prompt, get an output, maybe tweak and retry. But that’s not how creative escalation works.
Iteration: how humans make things funnier, weirder, more compelling. Stand-up comedians test a joke, exaggerate the punchline, add callbacks, escalate. Same principle here.
The “make it more ridiculous” prompt works because it’s contextual. ChatGPT understands “more” relative to what it just made. Faster and often more effective than rewriting a detailed prompt from scratch.
The DALL-E 3 Deadline You Need to Know
The AI model powering this entire workflow is about to change. DALL-E 3 is being deprecated on May 12, 2026 (OpenAI announced this November 14, 2025).
After mid-May, ChatGPT will use GPT Image 1.5 instead. The new model offers better editing, faster generation, improved consistency (according to OpenAI’s product docs). But the behavior of iterative prompts like “make it more ridiculous” might shift. If you’re relying on this technique for content creation, test it again after the switchover.
When It Breaks Down
| Problem | Why It Happens | Fix |
|---|---|---|
| Face becomes unrecognizable | Iterating on iterations loses original reference | Re-upload your photo and specify “keep this face” |
| ChatGPT ignores “more ridiculous” | Vague instruction after several rounds | Get specific: “add 10 more coffee mugs” or “make the background chaotic” |
| Image quality degrades | Compression artifacts compound over iterations | Download the best version early, restart from there if needed |
| Daily limit hits mid-session | Each iteration counts as a new image generation | Upgrade to ChatGPT Plus or pace your iterations |
Alternatives If You Hit the Wall
If ChatGPT’s daily limits or face-drift issues frustrate you, other tools handle iterative editing differently.
Midjourney lets you use the “remix” feature to iterate on images while maintaining more control over what changes. You’ll need to describe changes more explicitly, but you won’t lose the original composition as quickly.
Stable Diffusion (via interfaces like Automatic1111 or ComfyUI) gives you full control over iteration. You can lock certain elements (like your face) using ControlNet while letting other parts of the image escalate. The learning curve is steeper, but there are no daily limits if you run it locally.
Leonardo.AI offers a “Canvas” mode where you can selectively edit parts of an image. Useful if you want to make the background more ridiculous while keeping your face intact.
But ChatGPT’s conversational interface is still the easiest. No Discord commands. No Python libraries. You just talk to it.
What I Learned After 20 Iterations
Iteration 1-2: Fun. The absurdity escalates in ways that feel intentional.
Iteration 3-5: Peak chaos. This is where the best, most shareable images happen.
Iteration 6+: Diminishing returns. Things get weird but not funnier. Your face is gone. The composition breaks.
The sweet spot is 3-4 rounds. Enough to push past the generic caricature everyone else is posting. Not so far that you lose coherence.
One more thing: the images that performed best on social media weren’t the most extreme ones. They were the ones where the escalation was surprising but still legible. People want to see “you, but funnier” – not abstract chaos.
FAQ
Can I use “make it more ridiculous” with any AI image generator?
Works best in ChatGPT. The conversational context carries over. Midjourney and DALL-E (via API) don’t have the same memory – you’d need to describe the entire scene again each time. Stable Diffusion can do it via img2img workflows, but you’ll manually adjust settings.
Does ChatGPT remember my face across different chat sessions?
No. Start a new conversation and ChatGPT won’t have access to images from previous threads unless you re-upload. The iteration trick only works within a single conversation. Close the chat and come back later? You’ll need to upload your base image again to continue. I found this out when I tried to pick up where I left off the next day – had to start from scratch.
What happens if I say “make it less ridiculous” after going too far?
It works, but think of it like steering a moving car, not rewinding a video. ChatGPT generates a new image based on the previous one (not undoing edits), so you won’t get the exact version from two steps ago. You might dial back the coffee tower from 20 mugs to 10, but the overall composition will still be different from iteration 3. Save earlier versions as you go if you want to backtrack cleanly.
Now go make something that gets weirder with every round. Screenshot the best one before ChatGPT eats your face.