Context Big (rgthree)

Context Big is the Context node with more pockets. Same job - bundle a pile of connections into one wire so your graph stops looking like a plate of spaghetti - but it carries a much longer list of fields: not just model/clip/vae/conditioning/latent, but also your sampler settings, SDXL text pairs, a mask, and a control net. If your Context node keeps running out of slots for the things you want to thread through a workflow, this is the one you graduate to.

Why you'd reach for it

The plain Context node holds the essentials - model, clip, vae, positive, negative, latent, images, seed. That covers a basic pipeline. But a real production workflow often wants to carry more down the same wire: the sampler and scheduler so every stage agrees on them, the step budget and cfg, an SDXL prompt's separate g/l text, a mask for inpainting, a control net. Context Big has slots for all of that, so the entire configuration of a complex graph rides on a single connection.

The best part: Context and Context Big are backwards compatible. A bundle built on a small Context can flow into a Context Big and vice versa - the extra fields just come along as empty when they're not set. You can even convert between the two from the right-click menu. So you don't have to commit up front; start with Context and swap to Big when you need the extra slots, with nothing breaking.

Inputs and outputs

Every input is optional - you fill in what you have and leave the rest empty. Beyond the standard base_ctx, model, clip, vae, positive, negative, latent, images, and seed, Context Big adds a lot: steps, step_refiner, cfg, ckpt_name, sampler, scheduler, clip_width, clip_height, the four SDXL text fields text_pos_g / text_pos_l / text_neg_g / text_neg_l, a mask, and a control_net.

The outputs mirror all of it: a single CONTEXT wire (RGTHREE_CONTEXT) carrying the whole bundle, plus every field exploded out individually - MODEL, CLIP, VAE, POSITIVE, NEGATIVE, LATENT, IMAGE, SEED, STEPS, STEP_REFINER, CFG, CKPT_NAME, SAMPLER, SCHEDULER, CLIP_WIDTH, CLIP_HEIGHT, the four TEXT fields, MASK, and CONTROL_NET. Carry the CONTEXT wire for tidiness; tap the individual outputs wherever a specific node needs one piece.

The base_ctx trick is the same as on the regular Context: connect an existing bundle and this node passes it through, overriding only the fields you newly connect. That's how you thread one wire through many stages, updating a piece or two at each.

Installing it

Context Big ships in rgthree-comfy. Install via ComfyUI Manager (search rgthree's ComfyUI Nodes) or clone it:

cd ComfyUI/custom_nodes
git clone https://github.com/rgthree/rgthree-comfy.git

Restart ComfyUI. No models, no heavy dependencies - the pack is graph and UI code.

Common issues

The honest caveat: don't reach for Big before you need it. A screen full of unused slots is its own kind of clutter, and the whole point of the pack is legibility. If you're only carrying the basics, plain Context is cleaner. Because they interoperate, switching up later costs nothing.

Second, as with any context node, it only carries what you put in - an unfilled slot passes through as empty, silently. If a downstream stage gets None for its control net or mask, trace back and confirm you connected one. A Display Any tapped onto an exploded output is the quick way to see what's actually in the bundle.

And the pack-wide note: if the node renders wrong, check whether ComfyUI's Nodes 2.0 frontend is enabled. The Vue rewrite broke several rgthree nodes and the standing community workaround is to leave it off.