JoyLTX Optional Image (I2V on/off)

Text-to-video on LTX-2.5 and image-to-video are two different latent paths: T2V starts from empty noise, I2V encodes your picture as the first frame. The JoyLTX canvases wire both, and JoyLTX Optional Image is the little gate that decides which one you get. It's deliberately boring, and that's the point - it's the thing that stops you from rebuilding the graph every time you want to start a take from a photo.

The node has one real control: enabled, a BOOLEAN that defaults to off. Turn it on and the node passes your image through as the image output and sets the enabled output to true. Leave it off and it emits None and false - nothing flows, and the Take canvas' latent constructor sees no image and falls back to a true text-to-video render.

The interesting part is what happens upstream. The image input is declared lazy, so when enabled is off the node tells ComfyUI not to compute it at all. A Load Image feeding this node is skipped entirely - no file read, no decode, no wasted milliseconds. "Off" doesn't mean "ignored," it means "never ran," which is the difference between a toggle and a bypass. (In the source this is the check_lazy_status pattern returning ["image"] only when enabled, so only the needed inputs execute.)

On the Take canvas, the enabled output is what drives the empty-latent-vs-image-to-video switch, and the image output feeds the I2V conditioning. So the whole "use a reference first frame or not" decision collapses into one checkbox. The Multishot sampler has the same idea as an optional start_image slot - flip the gate on, load a picture, and shot 1 starts from it instead of from text.

Two things to know. First, if you enable the node but never connect an image, it prints a note and falls back to T2V rather than erroring - handy when you're prototyping. Second, this is a plumbing node, not a creative one: it doesn't touch your image beyond passing it through. The actual image-to-video conditioning is done by the sampler. If you want an end frame or mid keyframes, that's JoyLTX Keyframes; this one only ever does frame zero.

It ships with the ComfyUI-JoyLTX25 pack - no extra dependencies, nothing to download. If you're hand-building an LTX-2.5 workflow instead of using the pack's canvases, this is the cleanest way to get the T2V/I2V toggle without a pile of bypassed nodes.