Wan2.2 I2V A14B: animate a supplied still. Same two-expert schedule as the T2V graph, with one structural difference taken from ComfyUI 0.34.0's own video_wan2_2_14B_i2v template — WanImageToVideo produces the positive, negative AND latent that both samplers consume, so the start image conditions the clip rather than merely seeding it. That template leaves clip_vision_output UNLINKED, so no CLIP-vision tower is loaded here. Defaults are the template's: euler/simple, shift 5.0, 20 steps split at 10, CFG 3.5, 81 frames at 16 fps.
17Graph nodes
13Node classes
18Inputs
22Models
Tags: motionwan2.2image-to-videorunpod-serverlessvolume-backed
Inputs (18)#
The typed parameter surface callers bind when they request this workflow. Enum options and numeric bounds are the values the workflow document declares.
start_imageimagerequiredThe still to animate. Its aspect should match the chosen resolution; Wan does not letterbox, it stretches.
end_imageimageOptional final frame. Supplying one switches the graph to WanFirstLastFrameToVideo, which interpolates between the two stills instead of animating forward from one. EXPERIMENTAL: proved offline against ComfyUI's node signature, never rendered on a GPU.
promptstringrequireddefault the paper boat rocks and drifts forward, water rippling around itWhat should HAPPEN in the clip. For image-to-video, describe the motion rather than the scene — the scene is already in the image.
negative_promptstringdefault 色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走Wan's own published default negative prompt (Chinese, as shipped). Replacing it with an English translation measurably weakens it — the model was trained with this text.
seedintegerdefault -1Random seed (-1 = random). Both experts share it.
resolutionenumdefault 480p_landscape_832x480Frame size. 720p costs roughly 2.4x a 480p clip of the same length. The two match_image_aspect_* values derive width and height from the supplied start image instead, scaling its SHORT side to 480 or 720 and snapping both to a multiple of 16 — Wan does not letterbox, it stretches, so a mismatched canvas distorts the still.
options:
480p_landscape_832x480480p_portrait_480x832720p_landscape_1280x720720p_portrait_720x1280square_624x624match_image_aspect_480pmatch_image_aspect_720plengthintegerdefault 81min 5max 121Frames, and it must be 4n+1 — Wan's temporal compression plus the first frame. At 16 fps the default 81 is about 5 seconds. The executor refuses anything else before billing (motion-caps.ts).
fpsintegerdefault 16min 8max 30Frame rate written into the container. Wan 2.2 was trained at 16 fps; a higher number here plays the same frames faster, it does not generate more.
stepsintegerdefault 20min 4max 40Total sampling steps across BOTH experts. Overridden by speed_mode=lightx2v_4step, which forces 4.
boundary_ratiofloatdefault 0.5min 0.1max 0.9Where the high-noise expert hands over to the low-noise one, as a fraction of steps. The template splits 20 at 10 and 4 at 2 — both 0.5.
cfgfloatdefault 3.5min 1.0max 10.0Classifier-free guidance, applied to both experts. Overridden by speed_mode=lightx2v_4step, which forces 1.0.
shiftfloatdefault 5.0min 1.0max 12.0ModelSamplingSD3 flow shift on both experts. 5.0 is the template value.
speed_modeenumdefault qualitylightx2v_4step splices the I2V 4-step distilled LoRA onto BOTH experts and forces the template's lightning schedule (4 steps, split at 2, CFG 1.0), overriding steps/cfg/boundary_ratio. Note the I2V LoRAs are v1 files, not the v1.1 the T2V graph uses — that is what Comfy-Org publishes.
options:
qualitylightx2v_4stepsamplerenumdefault eulerSampler for both experts
options:
euleruni_pcdpmpp_2mschedulerenumdefault simpleSigma schedule for both experts
options:
simplebetamotion_presetenumdefault noneA documented phrase appended to the prompt (motion_map); `none` appends nothing at all. This is prompt guidance and nothing more — there is no camera control in this graph, and the model may ignore it.
options:
nonestatic cameraslow push inorbit lefthandhelddolly outloraenumdefault noneOptional Wan 2.2 LoRA pair (A.03.03/A.03.04). Wan 2.2 A14B is two experts, so every entry is two Civitai files: its high-noise file loads on the high-noise expert and its low-noise file on the low-noise expert, both at lora_strength, and an entry missing either file is refused at render rather than loaded by halves. Only pairs whose Civitai versions name Wan Video 2.2 I2V-A14B as their base are offered here. camera-tiltdown-undershot-v2 = the camera descends while tilting up to a low angle; camera-tiltup-overhead-v1.1 = the camera travels forward and rises to an overhead view; hip-sway-type1 = hip sway; hip-bump-dance = hip bump dance. Trigger phrases are NOT appended: each entry's phrases, as its Civitai versions list them in trainedWords, are in option_details (optionDetails in the catalog listing), and the prompt should use the one that fits the subject. Every pair rendered one live clip on 2026-09-14 (A.05.02; see proof_note). none adds no loader.
options:
nonecamera-tiltdown-undershot-v2camera-tiltup-overhead-v1.1hip-sway-type1hip-bump-dancensfw-posing-nude-v2bouncing-boobs-wan22twerking-i2vslow-twerklora_strengthfloatdefault 1min -2max 2LoRA strength on BOTH experts (LoraLoaderModelOnly.strength_model on the high-noise and the low-noise half alike). A slider LoRA is meant to be driven negative as well as positive; the motion LoRAs sit at 1.0.
ComfyUI node graph (17)#
The executable ComfyUI prompt graph: 17 nodes across 13 distinct node classes, wired by 22 data dependencies. Nodes tinted green come from a custom node pack this workflow declares; the rest are ComfyUI core / baked-community classes.
flowchart TD
n0["1: UNETLoader"]
n1["2: UNETLoader"]
n2["3: CLIPLoader"]
n3["4: VAELoader"]
n4["5: ModelSamplingSD3"]
n5["6: ModelSamplingSD3"]
n6["7: CLIPTextEncode"]
n7["8: CLIPTextEncode"]
n8["9: LoadImage"]
n9["10: KSamplerAdvanced"]
n10["11: KSamplerAdvanced"]
n11["12: VAEDecode"]
n12["13: CreateVideo"]
n13["14: SaveVideo"]
n14["15: ImageFromBatch"]
n15["16: SaveImage"]
n16["17: WanImageToVideo"]
n0 -->|model| n4
n1 -->|model| n5
n2 -->|clip| n6
n2 -->|clip| n7
n4 -->|model| n9
n16 -->|positive| n9
n16 -->|negative| n9
n16 -->|latent_image| n9
n5 -->|model| n10
n16 -->|positive| n10
n16 -->|negative| n10
n9 -->|latent_image| n10
n10 -->|samples| n11
n3 -->|vae| n11
n11 -->|images| n12
n12 -->|video| n13
n11 -->|image| n14
n14 -->|images| n15
n6 -->|positive| n16
n7 -->|negative| n16
n3 -->|vae| n16
n8 -->|start_image| n16
Nodes (17)#
1UNETLoadercoreunet_name = wan2.2_i2v_high_noise_14B_fp16.safetensorsweight_dtype = defaultoutputs:
MODEL2UNETLoadercoreunet_name = wan2.2_i2v_low_noise_14B_fp16.safetensorsweight_dtype = defaultoutputs:
MODEL3CLIPLoadercoreclip_name = umt5_xxl_fp16.safetensorstype = wandevice = defaultoutputs:
CLIP4VAELoadercorevae_name = wan_2.1_vae.safetensorsoutputs:
VAE5ModelSamplingSD3coremodel = ◂ node 1 · out[0]shift = {{shift}} tmploutputs:
MODEL6ModelSamplingSD3coremodel = ◂ node 2 · out[0]shift = {{shift}} tmploutputs:
MODEL7CLIPTextEncodecoretext = {{constructed_prompt}} tmplclip = ◂ node 3 · out[0]outputs:
CONDITIONING8CLIPTextEncodecoretext = {{negative_prompt}} tmplclip = ◂ node 3 · out[0]outputs:
CONDITIONING9LoadImagecoreimage = {{start_image}} tmploutputs:
IMAGEMASK10KSamplerAdvancedcoremodel = ◂ node 5 · out[0]positive = ◂ node 17 · out[0]negative = ◂ node 17 · out[1]latent_image = ◂ node 17 · out[2]add_noise = enablenoise_seed = {{seed}} tmplsteps = {{steps}} tmplcfg = {{cfg}} tmplsampler_name = {{sampler}} tmplscheduler = {{scheduler}} tmplstart_at_step = 0end_at_step = {{int(round(steps * boundary_ratio))}} tmplreturn_with_leftover_noise = enableoutputs:
LATENT11KSamplerAdvancedcoremodel = ◂ node 6 · out[0]positive = ◂ node 17 · out[0]negative = ◂ node 17 · out[1]latent_image = ◂ node 10 · out[0]add_noise = disablenoise_seed = {{seed}} tmplsteps = {{steps}} tmplcfg = {{cfg}} tmplsampler_name = {{sampler}} tmplscheduler = {{scheduler}} tmplstart_at_step = {{int(round(steps * boundary_ratio))}} tmplend_at_step = {{steps}} tmplreturn_with_leftover_noise = disableoutputs:
LATENT12VAEDecodecoresamples = ◂ node 11 · out[0]vae = ◂ node 4 · out[0]outputs:
IMAGE13CreateVideocoreimages = ◂ node 12 · out[0]fps = {{fps}} tmploutputs:
VIDEO14SaveVideocorevideo = ◂ node 13 · out[0]filename_prefix = wan22-i2vformat = mp4codec = h26415ImageFromBatchcoreimage = ◂ node 12 · out[0]batch_index = {{length - 1}} tmpllength = 1outputs:
IMAGE16SaveImagecoreimages = ◂ node 15 · out[0]filename_prefix = wan22-i2v-lastframe17WanImageToVideocorepositive = ◂ node 7 · out[0]negative = ◂ node 8 · out[0]vae = ◂ node 4 · out[0]start_image = ◂ node 9 · out[0]width = {{resolution_map[resolution].width}} tmplheight = {{resolution_map[resolution].height}} tmpllength = {{length}} tmplbatch_size = 1outputs:
CONDITIONINGCONDITIONINGLATENTPrompt construction#
template{base_prompt}{motion}Variables (2)#
base_prompt{{prompt}} tmplmotion{{motion_map[motion_preset]}} tmplParameter banks (7)#
The prompt / configuration lookup tables this workflow keys into from its inputs — the vocabulary that turns a style / palette / preset selection into graph parameters.
option_policy (1)#
lora{"camera-tiltdown-undershot-v2": {"distribution": "service", "content": "neutral"}, "camera-tiltup-overhead-v1.1": {"distribution": "service", "content": "neutral"}, "hip-sway-type1": {"distribution": "service", "content": "racy"}, "hip-bump-dance": {"distribution": "service", "content": "racy"}, "nsfw-posing-nude-v2": {"distribution": "internal_only", "content": "explicit"}, "bouncing-boobs-wan22": {"distribution": "internal_only", "content": "explicit"}, "twerking-i2v": {"distribution": "internal_only", "content": "explicit"}, "slow-twerk": {"distribution": "internal_only", "content": "explicit"}}resolution_map (7)#
480p_landscape_832x480{"width": 832, "height": 480}480p_portrait_480x832{"width": 480, "height": 832}720p_landscape_1280x720{"width": 1280, "height": 720}720p_portrait_720x1280{"width": 720, "height": 1280}square_624x624{"width": 624, "height": 624}match_image_aspect_480p{"width": 832, "height": 480}match_image_aspect_720p{"width": 1280, "height": 720}requires_families (2)#
wan22-i2vwan-sharedmotion_map (6)#
nonestatic camera, static cameraslow push in, slow push inorbit left, camera orbits lefthandheld, handheld cameradolly out, camera dollies outlora_pair_map (8)#
camera-tiltdown-undershot-v2{"high": "wan22-camera-tiltdown-undershot-v2-high.safetensors", "low": "wan22-camera-tiltdown-undershot-v2-low.safetensors"}camera-tiltup-overhead-v1.1{"high": "wan22-camera-tiltup-overhead-v1.1-high.safetensors", "low": "wan22-camera-tiltup-overhead-v1.1-low.safetensors"}hip-sway-type1{"high": "wan22-hip-sway-type1-high.safetensors", "low": "wan22-hip-sway-type1-low.safetensors"}hip-bump-dance{"high": "wan22-hip-bump-dance-high.safetensors", "low": "wan22-hip-bump-dance-low.safetensors"}nsfw-posing-nude-v2{"high": "wan22-nsfw-posing-nude-v2-high.safetensors", "low": "wan22-nsfw-posing-nude-v2-low.safetensors"}bouncing-boobs-wan22{"high": "wan22-bouncing-boobs-wan22-high.safetensors", "low": "wan22-bouncing-boobs-wan22-low.safetensors"}twerking-i2v{"high": "wan22-twerking-i2v-high.safetensors", "low": "wan22-twerking-i2v-low.safetensors"}slow-twerk{"high": "wan22-slow-twerk-high.safetensors", "low": "wan22-slow-twerk-low.safetensors"}post_render (3)#
wan_lora_pair{"input": "lora", "strength_input": "lora_strength", "pair_map": "lora_pair_map", "none_value": "none"}wan_speed_mode{"input": "speed_mode", "fast_value": "lightx2v_4step", "lora_high": "wan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensors", "lora_low": "wan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensors", "steps": 4, "split": 2, "cfg": 1.0, "strength": 1.0}wan_first_last_frame{"input": "end_image", "node_class": "WanImageToVideo", "swap_to": "WanFirstLastFrameToVideo"}submission (1)#
fit_to_image{"input": "resolution", "image_input": "start_image", "match_values": {"match_image_aspect_480p": 480, "match_image_aspect_720p": 720}, "node_classes": ["WanImageToVideo", "WanFirstLastFrameToVideo"], "multiple_of": 16, "_comment": "The resolution_map entries for these two values are the tier's landscape default and are only what the template renders before this transform overwrites them; the real numbers come from the image."}Models & dependencies#
Models required (22)#
wan2.2_i2v_high_noise_14B_fp16.safetensorswan2.2_i2v_low_noise_14B_fp16.safetensorsumt5_xxl_fp16.safetensorswan_2.1_vae.safetensorswan2.2_i2v_lightx2v_4steps_lora_v1_high_noise.safetensorswan2.2_i2v_lightx2v_4steps_lora_v1_low_noise.safetensorswan22-camera-tiltdown-undershot-v2-high.safetensorswan22-camera-tiltdown-undershot-v2-low.safetensorswan22-camera-tiltup-overhead-v1.1-high.safetensorswan22-camera-tiltup-overhead-v1.1-low.safetensorswan22-hip-sway-type1-high.safetensorswan22-hip-sway-type1-low.safetensorswan22-hip-bump-dance-high.safetensorswan22-hip-bump-dance-low.safetensorswan22-nsfw-posing-nude-v2-high.safetensorswan22-nsfw-posing-nude-v2-low.safetensorswan22-bouncing-boobs-wan22-high.safetensorswan22-bouncing-boobs-wan22-low.safetensorswan22-twerking-i2v-high.safetensorswan22-twerking-i2v-low.safetensorswan22-slow-twerk-high.safetensorswan22-slow-twerk-low.safetensorsOutput contract#
What a successful run of this workflow returns.
typevideoformatmp4codech264fps_sourcefixedaudiofalsealphafalseexpected_dimensions{"width": "{{resolution_map[resolution].width}}", "height": "{{resolution_map[resolution].height}}", "frames": "{{length}}"}Taxonomy & routing#
How the control plane classifies this workflow — from the committed workflow-taxonomy-registry.json. It drives the consistency / control surface the agentic director can exercise over the workflow.
assetFamilykeyframe-anchored-motionoutputPackageProfilevideo-master-profilecontrolModalitiesmodel-locksampler-scheduler-lockseed-lockprompt-template-lockkeyframe-anchor-lockmotion-vector-locktemporal-lockconsistencyDimensionsmotionidentitylightingenvironmentnotesWan2.2 I2V A14B. keyframe-anchor-lock is real here and not in the T2V graph: the start image is carried through WanImageToVideo into both experts' conditioning, so it anchors the clip rather than only seeding it.