I Asked Four AI Models to Age the Same Character - Here Are the Results

One of the more challenging tasks in AI image generation is aging a character while preserving their identity. In this experiment, I compare several leading image generation models to see which one produces the most convincing age progression. Base Image This is my base image of an 18-year-old woman. Prompt For every model, I used the same prompt: half body frontal shot of (25, 35, 45, 55) year old woman in elegant evening dress, medium length hair, looks at camera, white background Image Models Tested 1. Grok Imagine Quality 2. Wan 2.7 Pro 3. Seedream 5 4. GPT Image 2.0 Note: For easier comparison, I've included the original 18-year-old base image alongside each age group so you can see how well each model preserves the character's identity while aging her. Age 25 Base Image (above left), Seedream 5.0 (above middle) and Wan 2.7 Pro (above right) Grok Imagine Quality (above left) and GPT Image 2.0 (above right) Age 35 Base Image (above left), Seedream 5.0 (above middle) and Wan 2.7 Pro (above right) Grok Imagine Quality (above left) and GPT Image 2.0 (above right) Age 45 Base Image (above left), Seedream 5.0 (above middle) and Wan 2.7 Pro (above right) Grok Imagine Quality (above left) and GPT Image 2.0 (above right) Age 55 Base Image (above left), Seedream 5.0 (above middle) and Wan 2.7 Pro (above right) Grok Imagine Quality (above left) and GPT Image 2.0 (above right) Conclusion After comparing all four models, I found Seedream 5 to produce the most realistic aging overall. It consistently added age-related facial features while preserving the character's identity. Here is my personal ranking for this test: 1. Seedream 5 2. Wan 2.7 Pro 3. GPT Image 2.0 4. Grok Imagine Quality One thing I noticed is that GPT Image 2.0 and Grok Imagine Quality tend to make older characters look younger than the specified age. Their 45- and 55-year-old results often appeared closer to someone in their late 30s or 40s. Wan 2.7 Pro showed good aging capability and has a lot of potential, but it was less consistent. In several outputs, it changed the character's hairstyle, and occasionally the framing became more zoomed in than the original composition. Bonus Observation: Fashion Although Seedream 5 performed best at aging, I think GPT Image 2.0 generated the most stylish and natural-looking clothing. Its fashion choices generally felt more refined and believable. My Recommended Workflow If your goal is to create an older version of a character while maintaining high visual quality, this is the workflow I recommend: 1. Use Seedream 5 to age the character. 2. Import the result into GPT Image 2.0 to refine the clothing, hairstyle, or overall styling while preserving the aged appearance. This combination produced the best overall results in my testing. I hope you found this comparison helpful. If you've tried other image generation models for aging characters, I'd love to hear about your experience.