RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2089581532066443265?inviteCode=rh-v1111
Anima 2.9B Official Text-to-Image
Purpose: the baseline text-to-image graph for the Anima 2.9B preview checkpoint, using the official Qwen 0.6B text encoder and two selectable latent canvases. It is the reference file the other five workflows in this set build on.
Graph (as saved)
17 nodes / 21 links.
- UNETLoader:
Anima-2.9B-preview-v1.safetensors; CLIPLoader:qwen_3_06b_base.safetensorswith typestable_diffusion; VAELoaderKJ:qwen_image_vae.safetensors(main_device, fp32). - Sampler:
SamplerCustomAdvanced+KSamplerSelecteuler +BasicSchedulersgm_uniform, 50 steps, denoise 1;CFGGuiderCFG 8. - A separate rgthree "KSampler Config" node stores 28 steps / CFG 5 /
res_multistep/linear_quadraticas a reference config; the active sampling chain is the SamplerCustomAdvanced/SGM-uniform path above. - Two EmptyLatentImage canvases (1216x832 and 1152x1536) switch through an ImpactSwitch. SaveImage prefix
Anima_Sample/preview_2.9b. - Positive prompt is the author's demo tag block for a Genshin-style two-character scene; negative is the official 2.9B baseline quality-negative block. Both are demo content.
How to use
Replace the prompt and pick the canvas, or edit the two EmptyLatent sizes. Keep the euler/SGM-uniform 50-step path unless you deliberately switch to the stored rgthree config. The official checkpoint is a preview build; treat outputs as such.
Limits and not-verified items
No inference was run for this listing: no speed, VRAM or quality claim is made. Nodes, prompts and canvas sizes are quoted as saved. Weights, VAE, encoder and demo images are not bundled; check their licenses separately. The cover image is a promotional still from the source video, 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.
Description
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2089581532066443265?inviteCode=rh-v1111
Anima 2.9B Official Text-to-ImagePurpose: the baseline text-to-image graph for the Anima 2.9B preview checkpoint, using the official Qwen 0.6B text encoder and two selectable latent canvases. It is the reference file the other five workflows in this set build on.
Graph (as saved)17 nodes / 21 links.
- UNETLoader:
Anima-2.9B-preview-v1.safetensors; CLIPLoader:qwen_3_06b_base.safetensorswith typestable_diffusion; VAELoaderKJ:qwen_image_vae.safetensors(main_device, fp32). - Sampler:
SamplerCustomAdvanced+KSamplerSelecteuler +BasicSchedulersgm_uniform, 50 steps, denoise 1;CFGGuiderCFG 8. - A separate rgthree "KSampler Config" node stores 28 steps / CFG 5 /
res_multistep/linear_quadraticas a reference config; the active sampling chain is the SamplerCustomAdvanced/SGM-uniform path above. - Two EmptyLatentImage canvases (1216x832 and 1152x1536) switch through an ImpactSwitch. SaveImage prefix
Anima_Sample/preview_2.9b. - Positive prompt is the author's demo tag block for a Genshin-style two-character scene; negative is the official 2.9B baseline quality-negative block. Both are demo content.
Replace the prompt and pick the canvas, or edit the two EmptyLatent sizes. Keep the euler/SGM-uniform 50-step path unless you deliberately switch to the stored rgthree config. The official checkpoint is a preview build; treat outputs as such.
Limits and not-verified itemsNo inference was run for this listing: no speed, VRAM or quality claim is made. Nodes, prompts and canvas sizes are quoted as saved. Weights, VAE, encoder and demo images are not bundled; check their licenses separately. The cover image is a promotional still from the source video, 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.