IMAGE MODEL COMPARISON / 06 NOV 2026
What stayed true
to the garment?
The same eight mannequin photos were tested against a new image model. Compare the source, the previously accepted Scalio HD V1 result, and the new GPT Image 2.6 result. Click any photo to inspect the full-size file.
The new run makes chokers, choli handwork, shoulder details and bead strands easier to see, with a quieter background. The accepted HD V1 set has more varied editorial staging. The new set repeats a similar front-facing pose and staircase composition.
ABOUT THIS EVAL
A visual comparison, not a controlled benchmark.
Each GPT Image 2.6 edit received the original mannequin photo and an outfit-specific brief for a standing female model, faithful garment details, and a subdued blue-hour garden staircase. The HD V1 output was used only for comparison, never as an input to the new model. One new result was generated per source image. These new results have not been approved by the user.
The new files are 1024 × 1536. The accepted HD V1 files are 1792 × 2400, so inspect the source and full-size images before judging fine embroidery or print quality.