25Graph nodes
22Node classes
12Inputs
14Models
Tags: chromarunpod-serverlessvolume-backedtext-to-image
Inputs (12)#
The typed parameter surface callers bind when they request this workflow. Enum options and numeric bounds are the values the workflow document declares.
promptstringrequireddefault a close-up photograph of a tiger's eye, intricately detailed fur, natural lighting, sharp focus, highest qualityPositive prompt in natural prose (Chroma responds to descriptive photography language)
negative_promptstringdefault This greyscale unfinished sketch has bad proportions, is featureless and disfigured. It is a blurry ugly mess with excessive gaussian blur, riddled with watermarks and signatures, smudged with leaking colors and nonsensical orientation of objects. Messy and abstract, filled with artifacts that disrupt the coherency of the composition, extreme chromatic aberrations and inconsistent lighting. Dull, monochrome colors and countless artistic errors.Negative prompt (Chroma is CFG-guided, so this is honoured)
checkpointenumdefault chroma1-hdFull-precision Chroma checkpoint on the network volume. ungloryhail-bf16 = UnGloryHail BF16 (A.03.01, Civitai model-version 2898818, the BF16 file 2776781 rather than the version's fp8 primary; content explicit; not rendered on a GPU yet, A.03.01 live still owed).
options:
chroma1-hduncanny-v13-bf16photonic-fusion-debloated-bf16ungloryhail-bf16stepsintegerdefault 26min 8max 60Sampling steps (26 is the Chroma1-HD reference)
cfgfloatdefault 3.8min 1.0max 10.0Classifier-free guidance (3.8 is the Chroma1-HD reference)
seedintegerdefault -1Random seed (-1 = random)
aspectenumdefault square_1024Output aspect preset
options:
square_1024portrait_896x1152landscape_1152x896portrait_832x1216landscape_1216x832batch_sizeintegerdefault 1min 1max 1Images per run. Pinned to 1: the executor's resolution-scaled batch cap (C.12.19) allows a second latent only below ~0.5 MP, and every aspect preset here is ~1 MP. batch_size 2 at 896x1152 killed a live worker.
samplerenumdefault eulerSampler. euler is the Chroma1-HD reference; the rest are samplers already in use elsewhere in this catalog.
options:
eulereuler_ancestraldpmpp_2mdpmpp_2m_sdeuni_pcres_multistepupscaleenumdefault noneOptional still-upscale tail, 2x. seedvr2_3b (A.01.08) resizes the still 2x with lanczos and restores it in one step with SeedVR2 3B (fp16) on ComfyUI's core SeedVR2 nodes, as the SeedVR2 3B image template does at 4x; its first use loads the seedvr2-3b family (7.29 GB) into the model cache. Only 2x is offered: the worker returns the PNG inline in the job output, and a 4x (~16 MP) still would not fit RunPod's job-output ceiling until outputs travel by the volume (V.12.05). none leaves the graph exactly as before. The Real-ESRGAN x2 tail (C.12.02) was retired once SeedVR2 3B had rendered (A.01.08); the enum refuses a job that still names x2.
options:
noneseedvr2_3bloraenumdefault noneOptional LoRA on the Chroma model (C.12.01; four picked by the user on 2026-09-13, all from CivitAI with every commercial flag). lenovo-ultrareal = photorealism enhancer (its trigger word `l3n0v0` is appended to the prompt automatically); chroma-realistic-slider = realism up/down by strength; chroma-defluxer = removes the residual Flux look; style-muk = illustration style. A.03.01 adds two more, also with every commercial flag: nicegirls-ultrareal = NiceGirls UltraReal Chroma (model-version 2487473, content racy); nsfw-realism-v1 = NSFW Realism v1.0 (model-version 2357084, content explicit). Neither version lists trained words, so neither appends a trigger, and neither has rendered on a GPU yet (A.03.01 live stills owed). none leaves the graph exactly as before.
options:
nonelenovo-ultrarealchroma-realistic-sliderchroma-defluxerstyle-muknicegirls-ultrarealnsfw-realism-v1lora_strengthfloatdefault 1.0min -2.0max 2.0LoRA strength on the model (LoraLoaderModelOnly.strength_model). The slider LoRA is meant to be driven negative as well as positive; the others sit at 1.0.
ComfyUI node graph (25)#
The executable ComfyUI prompt graph: 25 nodes across 22 distinct node classes, wired by 30 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: CLIPLoader"]
n2["3: VAELoader"]
n3["4: ModelSamplingAuraFlow"]
n4["5: T5TokenizerOptions"]
n5["6: CLIPTextEncode"]
n6["7: CLIPTextEncode"]
n7["8: CFGGuider"]
n8["9: KSamplerSelect"]
n9["10: BetaSamplingScheduler"]
n10["11: RandomNoise"]
n11["12: EmptySD3LatentImage"]
n12["13: SamplerCustomAdvanced"]
n13["14: VAEDecode"]
n14["15: SaveImage"]
n15["19: LoraLoaderModelOnly"]
n16["20: UNETLoader"]
n17["21: VAELoader"]
n18["22: ResizeImageMaskNode"]
n19["23: SeedVR2Preprocess"]
n20["24: VAEEncodeTiled"]
n21["25: SeedVR2Conditioning"]
n22["26: KSampler"]
n23["27: VAEDecodeTiled"]
n24["28: SeedVR2PostProcessing"]
n15 -->|model| n3
n1 -->|clip| n4
n4 -->|clip| n5
n4 -->|clip| n6
n3 -->|model| n7
n5 -->|positive| n7
n6 -->|negative| n7
n3 -->|model| n9
n10 -->|noise| n12
n7 -->|guider| n12
n8 -->|sampler| n12
n9 -->|sigmas| n12
n11 -->|latent_image| n12
n12 -->|samples| n13
n2 -->|vae| n13
n0 -->|model| n15
n13 -->|input| n18
n18 -->|resized_images| n19
n19 -->|pixels| n20
n17 -->|vae| n20
n16 -->|model| n21
n20 -->|vae_conditioning| n21
n16 -->|model| n22
n21 -->|positive| n22
n21 -->|negative| n22
n20 -->|latent_image| n22
n22 -->|samples| n23
n17 -->|vae| n23
n23 -->|images| n24
n18 -->|original_resized_images| n24
Nodes (25)#
1UNETLoadercoreunet_name = {{checkpoint_map[checkpoint]}} tmplweight_dtype = defaultoutputs:
MODEL2CLIPLoadercoreclip_name = t5xxl_fp16.safetensorstype = chromadevice = defaultoutputs:
CLIP3VAELoadercorevae_name = ae.safetensorsoutputs:
VAE4ModelSamplingAuraFlowcoremodel = ◂ node 19 · out[0]shift = 1.0outputs:
MODEL5T5TokenizerOptionscoreclip = ◂ node 2 · out[0]min_padding = 1min_length = 0outputs:
CLIP6CLIPTextEncodecoretext = {{constructed_prompt}}{{lora_trigger_map[lora]}} tmplclip = ◂ node 5 · out[0]outputs:
CONDITIONING7CLIPTextEncodecoretext = {{negative_prompt}} tmplclip = ◂ node 5 · out[0]outputs:
CONDITIONING8CFGGuidercoremodel = ◂ node 4 · out[0]positive = ◂ node 6 · out[0]negative = ◂ node 7 · out[0]cfg = {{cfg}} tmploutputs:
GUIDER9KSamplerSelectcoresampler_name = {{sampler}} tmploutputs:
SAMPLER10BetaSamplingSchedulercoremodel = ◂ node 4 · out[0]steps = {{steps}} tmplalpha = 0.45beta = 0.45outputs:
SIGMAS11RandomNoisecorenoise_seed = {{seed}} tmploutputs:
NOISE12EmptySD3LatentImagecorewidth = {{aspect_map[aspect].width}} tmplheight = {{aspect_map[aspect].height}} tmplbatch_size = {{batch_size}} tmploutputs:
LATENT13SamplerCustomAdvancedcorenoise = ◂ node 11 · out[0]guider = ◂ node 8 · out[0]sampler = ◂ node 9 · out[0]sigmas = ◂ node 10 · out[0]latent_image = ◂ node 12 · out[0]outputs:
LATENTLATENT14VAEDecodecoresamples = ◂ node 13 · out[0]vae = ◂ node 3 · out[0]outputs:
IMAGE15SaveImagecoreimages = {{upscale_output_map[upscale]}} tmplfilename_prefix = isis/chroma-txt2img19LoraLoaderModelOnlycoremodel = ◂ node 1 · out[0]lora_name = {{lora_map[lora]}} tmplstrength_model = {{lora_strength}} tmploutputs:
MODEL20UNETLoadercoreunet_name = seedvr2_3b_fp16.safetensorsweight_dtype = defaultoutputs:
MODEL21VAELoadercorevae_name = seedvr2_ema_vae_fp16.safetensorsoutputs:
VAE22ResizeImageMaskNodecoreinput = ◂ node 14 · out[0]resize_type = scale by multiplierresize_type.multiplier = 2.0scale_method = lanczosoutputs:
IMAGE23SeedVR2Preprocesscoreresized_images = ◂ node 22 · out[0]outputs:
IMAGE24VAEEncodeTiledcorepixels = ◂ node 23 · out[0]vae = ◂ node 21 · out[0]tile_size = 512overlap = 128temporal_size = 4096temporal_overlap = 8outputs:
LATENT25SeedVR2Conditioningcoremodel = ◂ node 20 · out[0]vae_conditioning = ◂ node 24 · out[0]outputs:
CONDITIONINGCONDITIONING26KSamplercoremodel = ◂ node 20 · out[0]positive = ◂ node 25 · out[0]negative = ◂ node 25 · out[1]latent_image = ◂ node 24 · out[0]seed = {{seed}} tmplsteps = 1cfg = 1.0sampler_name = eulerscheduler = simpledenoise = 1.0outputs:
LATENT27VAEDecodeTiledcoresamples = ◂ node 26 · out[0]vae = ◂ node 21 · out[0]tile_size = 512overlap = 128temporal_size = 4096temporal_overlap = 8outputs:
IMAGE28SeedVR2PostProcessingcoreimages = ◂ node 27 · out[0]original_resized_images = ◂ node 22 · out[0]color_correction_method = noneoutputs:
IMAGEPrompt construction#
template{base_prompt}Variables (1)#
base_prompt{{prompt}} 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 (2)#
checkpoint{"chroma1-hd": {"distribution": "service", "content": "neutral"}, "uncanny-v13-bf16": {"distribution": "service", "content": "neutral"}, "photonic-fusion-debloated-bf16": {"distribution": "internal_only", "content": "neutral"}, "ungloryhail-bf16": {"distribution": "service", "content": "explicit"}}lora{"lenovo-ultrareal": {"distribution": "service", "content": "neutral"}, "chroma-realistic-slider": {"distribution": "service", "content": "neutral"}, "chroma-defluxer": {"distribution": "service", "content": "neutral"}, "style-muk": {"distribution": "service", "content": "neutral"}, "nicegirls-ultrareal": {"distribution": "service", "content": "racy"}, "nsfw-realism-v1": {"distribution": "service", "content": "explicit"}}checkpoint_map (4)#
chroma1-hdChroma1-HD.safetensorsuncanny-v13-bf16uncannyPhotorealism_v13_bf16.safetensorsphotonic-fusion-debloated-bf16photonicFusionChroma_debloated_bf16.safetensorsungloryhail-bf16ungloryhail_bf16.safetensorsaspect_map (5)#
square_1024{"width": 1024, "height": 1024}portrait_896x1152{"width": 896, "height": 1152}landscape_1152x896{"width": 1152, "height": 896}portrait_832x1216{"width": 832, "height": 1216}landscape_1216x832{"width": 1216, "height": 832}requires_families (2)#
chroma-coreseedvr2-3blora_map (7)#
nonelenovo-ultrarealchroma-lenovo-ultrareal.safetensorschroma-realistic-sliderchroma-realistic-slider.safetensorschroma-defluxerchroma-defluxer.safetensorsstyle-mukchroma-style-muk.safetensorsnicegirls-ultrarealchroma-nicegirls-ultrareal.safetensorsnsfw-realism-v1chroma-nsfw-realism-v1.safetensorslora_trigger_map (7)#
nonelenovo-ultrareal, l3n0v0chroma-realistic-sliderchroma-defluxerstyle-muknicegirls-ultrarealnsfw-realism-v1upscale_output_map (2)#
none140seedvr2_3b280Models & dependencies#
Models required (14)#
Chroma1-HD.safetensorsuncannyPhotorealism_v13_bf16.safetensorsphotonicFusionChroma_debloated_bf16.safetensorsungloryhail_bf16.safetensorst5xxl_fp16.safetensorsae.safetensorschroma-lenovo-ultrareal.safetensorschroma-realistic-slider.safetensorschroma-defluxer.safetensorschroma-style-muk.safetensorschroma-nicegirls-ultrareal.safetensorschroma-nsfw-realism-v1.safetensorsseedvr2_3b_fp16.safetensorsseedvr2_ema_vae_fp16.safetensorsOutput contract#
What a successful run of this workflow returns.
typeimageformatpngTaxonomy & 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.
assetFamilyportrait-hero-image-bundleoutputPackageProfileimage-single-profilecontrolModalitiesmodel-locksampler-scheduler-lockseed-lockprompt-template-lockconsistencyDimensionsidentitylightinglensnotesChroma full-precision text-to-image on RunPod serverless.