# Marigold V2 Turns a Diffusion Transformer Into a Single-Step Depth Predictor

Huawei's Bayer Lab reports 16 to 26 percent lower relative depth error on KITTI and ETH3D, with fine-tuning that takes under a week on one consumer graphics card and Apache 2.0 weights.

- Published: 2026-09-12T06:23:24.914Z
- Canonical: https://polylog.news/ai/2026-09-12/marigold-v2-turns-a-diffusion-transformer-into-a-single-step
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [arXiv](https://arxiv.org/abs/2609.08084), [Hugging Face](https://huggingface.co/huawei-bayerlab/marigold-v2-0)

Marigold V2, from Huawei's Bayer Lab, is a fine-tuning protocol that converts a pretrained diffusion transformer into a single-step dense predictor for monocular depth, see-through depth, surface normals, and albedo. The single-step part is…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-09-12/marigold-v2-turns-a-diffusion-transformer-into-a-single-step (subscription information: https://polylog.news/pricing).