Breaking

Higgsfield Tests GPT-Image-2.5 Lighting Consistency for Stop Motion

September 9, 2026 at 01:52 EDT

Media Generation
Foundation Models

Higgsfield AI is testing GPT-Image-2.5 on a stop-motion-style sequence, focusing on whether successive generated frames can preserve consistent illumination. The experiment turns a core production problem—visual continuity—into an early practical test for OpenAI’s new image-generation release. OpenAI’s launch details show that ChatGPT Images 2.5 launched on September 8, 2026, alongside two API models: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.

SEPT. 8, 2026 · OPENAI

GPT-Image-2.5 Meets a Frame-Consistency Test

Higgsfield AI is testing whether generated stills can keep illumination consistent across a stop-motion-style sequence. The models are built for controlled image generation, but they do not output video.

The signal in the first tests

50%

OpenAI-reported latency reduction

API model variants

Generations in the matched test

9.2%

Flare pixel drift after turn 3

Matched latency test

1,413 output tokens · mean seconds

37.3s

GPT-Image-2

27.7s

Sunburst

19.7s

Flare

Shorter is faster. The comparison comes from a 41-generation test.

What looks promising

Flare was fastest in the matched test
OpenAI reports stronger multi-turn consistency
Lighting and texture are explicit quality targets

What remains open

Practical tests still found edit spillover
The API returns images, not video
No frame-stability score is published

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.