Category guide

Buy & sell ComfyUI workflows — portraits, relight, upscale, and more

A ComfyUI workflow is a saved node graph: loaders, samplers, ControlNet, IPAdapter, upscalers, and save nodes wired together. On Truve you can buy ComfyUI workflows for real production tasks (e.g. buy ComfyUI workflow portrait, relighting, product shots) or sell ComfyUI workflows you have already battle-tested — paid in USDC on Base, with encrypted delivery and a 5% platform fee (seller keeps 95%). This guide explains what makes a workflow listing trustworthy, how payments and file delivery work at launch, and how Truve fits next to forums, Patreon drops, and model hubs — written for both first-time buyers and creators who want Google to surface answers to long-tail queries like ComfyUI workflow marketplace and sell ComfyUI graph.

Why ComfyUI workflows are a real product category

Generative image pipelines stopped being a novelty for many studios the moment they had to ship consistent results under deadlines. A single screenshot of someone else’s graph is rarely enough: hidden group nodes, obscure preprocessor settings, and “works on my machine” VRAM assumptions waste hours. A packaged workflow — with notes, example outputs, and explicit dependency lists — is closer to consulting in a file than to a random preset. That is why search volume exists around phrases like buy ComfyUI workflow portrait, ComfyUI upscale workflow download, and batch ComfyUI workflow ecommerce: practitioners want reproducibility, not inspiration.

Truve treats workflows as first-class digital goods, the same way stock asset sites treat LUT packs or 3D rigs. Sellers name their price; buyers pay in stablecoin on a low-fee L2; the platform takes a transparent 5% fee so creators keep the majority of revenue. The goal is not to replace community knowledge — it is to put discovery, payment, and delivery in one place so commercial buyers can justify a purchase to their finance team instead of sending PayPal to a random Discord handle.

Buyer’s guide: what to check before you pay

Before you buy a ComfyUI workflow, read the listing the way you would read a software dependency manifest. Confirm the base model family (SD1.5, SDXL, Flux, specialized checkpoints) and whether the seller expects you to supply weights yourself — which is normal — or bundles small auxiliary files they actually have rights to ship. Check every custom node name and whether the graph relies on bleeding-edge nightly builds; if so, decide whether your studio policy allows that risk profile.

Ask yourself what resolution and aspect ratio you need for final delivery. A portrait pipeline tuned at 832×1216 may fall apart at double resolution unless the listing says otherwise. Look for VRAM guidance: a workflow that chains four ControlNets and a refiner is not wrong, but it is not for a 8 GB laptop GPU. Finally, scan preview images for watermarks or celebrity likenesses; ethical sellers use original or licensed reference subjects so you are not inheriting hidden IP problems in your client work.

Seller’s guide: listings that convert and stay supported

If you want to sell ComfyUI workflows, your competitive advantage is clarity under maintenance. Start with a one-paragraph promise: who the graph is for (solo portrait retouchers, Amazon sellers, anime colorists) and what outcome they get in measurable terms — for example “one-click relight from single JPEG to studio key at 2048px.” Follow with a dependency table: checkpoint filenames as published on Civitai or Hugging Face, LoRA names and recommended strengths, and each custom node repository with a commit hash you tested.

Include a first-run checklist buyers can follow in order: install nodes, place models, open workflow, fix two known path quirks, run default seed, compare to your reference PNG. Screenshots of the node group layout help people orient before they even load the file. State your support policy plainly: do you answer questions for 30 days, do you offer free graph updates when a major node renames inputs, or is the sale strictly as-is? Ambiguity creates chargebacks and bad reviews; explicit boundaries create professional repeat customers.

Models, custom nodes, and ComfyUI version ranges

Most disputes in workflow sales come from version skew. Pin your listing to a ComfyUI commit or release tag when possible, and list a Python and torch band if your nodes depend on CUDA extensions. If a popular node renames a socket between versions, document both names or ship two workflow files. For LoRA and textual inversion stacks, specify whether buyers should expect to merge prompts manually or use your baked group defaults.

When your workflow uses paid or non-redistributable checkpoints, never bundle the weights without permission. The industry-standard approach is: sell the graph only, link to official purchase pages, and optionally include a tiny placeholder checkpoint for layout testing if licenses allow. Truve is built for honest listings; misrepresenting what ships inside the ZIP undermines the whole category.

VRAM, inference speed, and batch generation for agencies

Agency buyers often need folder-in folder-out automation: thousands of SKUs with identical lighting rules. Describe how your graph expects batching — external script, ComfyUI’s queue, or an API payload pattern — and what breaks if filenames are not UTF-8 clean. Mention realistic seconds per image on a named GPU so art directors can estimate overnight render farms. If your workflow uses latent upscaling chains, note the VRAM spike during the upscale phase separately from base diffusion, because that is where many machines OOM even when base passes seem fine.

How checkout and file delivery work on Truve

Truve is a decentralized marketplace on Base. Buyers connect a self-custody wallet, pay in USDC, and receive encrypted payloads unlocked by smart-contract rules combined with Lit Protocol for access control — so files are not sitting on a public CDN link before payment. Sellers upload workflow bundles (JSON graphs, small helper images, README) within the platform’s size limits at launch; after a successful purchase, the buyer downloads through the same account-linked wallet flow described on the homepage. Fee structure is simple: 5% to the protocol, 95% to the creator, which compares favorably to double-digit cuts on many centralized creator platforms.

Compared to Discord sales, Patreon, and model repositories

Discord excels at fast feedback loops but is weak for SEO, search, and accounting. Patreon fits subscriptions and community tiers, yet many buyers only want a single graph, not a recurring bill. Model hubs focus on weights, not on end-to-end graphs with business logic. Truve sits in the gap: one-off commerce for sophisticated files, with crypto rails that work globally without card-network friction for digital goods. You can still cross-post; just keep pricing and license text consistent so buyers trust you everywhere.

Related on Truve: combine workflows with custom nodes & extensions when you ship code, or with prompt packs when your value is mostly text tokens rather than wiring.

What buyers get

Instead of rebuilding a graph from Discord screenshots, you download a ready-to-run ComfyUI workflow: drop the JSON (or packaged folder) into ComfyUI, fix paths to your checkpoints if needed, and run. Good listings include: required models, ComfyUI version, custom nodes, and example before/after images.

Popular niches (long-tail examples)

These are the kinds of searches power users type — and the workflows Truve is built to host:

• Portrait & beauty — skin-aware detail, face restore, consistent identity across seeds.
• Relight & studio — single-image relighting, fake studio setups, product-on-grey.
• Background & segmentation — subject cutout, background swap, inpaint cleanup.
• Upscale & print — 2×/4× chains, artifact control for print-ready stills.
• Batch & CSV — folder in, folder out; same graph for hundreds of SKUs.

Example listing ideas

Why sell ComfyUI workflows on Truve (not only in Discord)

Discord is great for community; it is a weak ComfyUI workflow marketplace: no search, no reviews, payments in DMs. Truve adds a catalog, on-chain payment, and automatic file delivery so people who search “sell ComfyUI workflow” or “ComfyUI workflow download” land on a serious storefront.

File size & safety at launch

At launch, Truve targets bundles under 50 MB per listing (graph + small assets). Always disclose third-party checkpoints and licenses in your description — buyers care that your ComfyUI workflow download is legal to use commercially.

If you need to ship larger auxiliary data (high-res mask packs, small video loops for I2V tests), split into a core workflow ZIP on Truve and optionally link buyers to a separate storage product you control — but never hide paid dependencies inside an opaque binary. Transparency is what turns occasional Google visitors into five-star reviews.

Glossary: terms people search alongside “ComfyUI workflow”

ControlNet adds spatial guidance (edges, depth, pose) so diffusion respects structure. IPAdapter injects reference images into cross-attention for style or identity consistency. KSampler is the default diffusion loop node; CFG and scheduler names in your graph should match what buyers see in their UI. Latent upscale means enlarging in latent space before final VAE decode, often sharper than pure pixel upscale but heavier on VRAM. Face detailer subgraphs run a second pass on detected faces. When your listing uses these words accurately, you help both humans and search engines understand fit.

FAQ — ComfyUI workflows on Truve