RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2089343554559758337?inviteCode=rh-v1111

Anima 2.9B 1.5x High-Resolution Upscale

Purpose: a two-pass Anima 2.9B generation that produces a 1 MP image, upscales the latent by 1.5x, and runs a low-denoise second pass to resolve detail. It is the simplest of the six files and the easiest one to tune.

Graph (as saved)

13 nodes / 17 links.

• UNETLoader Anima-2.9B-preview-v1.safetensors; CLIPLoader qwen_3_06b_base.safetensors (type stable_diffusion); VAELoaderKJ qwen_image_vae.safetensors.
• Base canvas: EmptyLatentImage 1024x1024, with a FluxResolutionNode at 1.0 MP / 3:4 style / 64 multiple.
• First KSampler: 28 steps, CFG 5, res_multistep, linear_quadratic, denoise 1.
• Then LatentUpscaleBy nearest-exact at 1.5x and a second KSampler: same 28 steps / CFG 5 / sampler / scheduler, denoise 0.35.
• SaveImage prefix Anima_2.9B/1.5x_hires.

How to use

Swap the prompt and canvas, keep the 0.35 second-pass denoise for a detail pass, or lower it if the upscale drifts. The 1.5x factor is applied on the latent, not on the decoded image, so the second pass sees the enlarged latent directly.

Limits and not-verified items

No inference was run for this listing: no speed, VRAM or quality claim is made; the saved values are quoted as-is. Weights, VAE, encoder and demo images are not bundled. The cover image is a promotional still, not a frame produced by this graph.

Tutorial: YouTube · Bilibili. Workflows supplied through AI-KSK; verify workflow, node, model, encoder, image and cover rights separately. No blanket commercial or redistribution license is asserted.