By Cemhan Biricik, Founder, ZSky AI
I have aphantasia. I cannot voluntarily picture images in my mind. For most of my life, my creative ideas existed only as feelings, structures, and words — never as pictures. I assumed I had no art in me.
Then I survived a traumatic brain injury that took my words for almost a year. Photography rebuilt the neural pathways the injury had broken. Neuroplasticity through creative work is documented in the literature. The camera was my therapy.
When AI image generation matured, I built ZSky AI — a free AI image and video generator running on 7 NVIDIA RTX 5090 GPUs in my US workstation — because I needed it for myself. Here's what surprised me about the experience.
You don't need to picture the image to make it.
The first thing I learned is that prompting is fundamentally a language task, not a visualization task. When I describe "a cinematic portrait of a young woman in a Brooklyn loft at golden hour, 35mm film grain, contemplative expression," I'm not picturing it. I'm composing words that have visual meaning attached. The image arrives as a surprise to me, every single time. I don't know what I'm going to see until I see it.
For most aphantasics I've talked to since launching ZSky, this is the experience too. We don't lose by having no mental image. We gain by having no mental image to be disappointed against. The output is what it is. We respond to it on its own terms.
Iteration matters more than first-shot accuracy.
Because I can't picture the target, I can't tell whether my first generation is "right." I have to generate, react, and re-prompt. This loop matters more for me than for neurotypical creators. Speed of iteration is the single most important UX axis for accessibility.
That's why I built ZSky to generate images in about 2 seconds and 1080p video in about 30 seconds. Anything slower would have made the iteration loop unusable for me. The free tier is permanent because I don't want anyone with aphantasia to pay for the privilege of iteration.
Reference images are critical.
When I want to make something specific — like a particular kind of room, lighting, or pose — I cannot describe it from memory because I have no visual memory. I have to upload a reference. ZSky's image-to-image feature is the most-used feature in my own workflow. I think this is true for most aphantasic creators.
If you're building an AI image tool, prioritize the image-to-image flow over text-to-image polish. People with aphantasia, low vision, or visual cortex damage are the canary for this UX choice — the feature is critical for us in a way that's invisible to neurotypical users until you ask.
The community is bigger than I knew.
When I launched ZSky, I assumed I was building a niche tool for a small population. The Aphantasia Network has 84,000 members worldwide. There are an estimated 12-40 million people with aphantasia. The TBI population is millions more. The neurodivergent population is much larger still. The "people who experience visual creativity differently" market is not niche — it's just been ignored.
I built the 1 Million Minds Eye Initiative.
ZSky now offers free lifetime Ultra tier access to 1,000,000 people with aphantasia, traumatic brain injury, visual cortex damage, or any condition that affects how you experience visual creation. No payment, no medical documentation, no expiration. Honor system.
It's named after ElevenLabs' 1 Million Voices Initiative for the same reason: AI as accessibility is the most important application of generative AI in this decade, and the first companies to embrace this framing will define how the rest of the industry positions itself.
The unfair advantage I didn't expect.
Building an AI image tool with aphantasia turns out to be an asset, not a disadvantage. I cannot dismiss outputs based on "this isn't what I imagined" because I never imagined them. I evaluate everything on its own merits. I'm a more honest user than someone who can compare against a mental reference.
If you have aphantasia and you're considering building or using AI image tools, the answer is unambiguously yes. The technology was built for people who can already see what they want. We get to be the people who see what we couldn't see before.
Try ZSky AI free →
1 Million Minds Eye Initiative →
Accessibility hub →
— Cemhan Biricik
Founder, ZSky AI
Top comments (0)