EasyAnimate Sampler for Text to Video

EasyAnimateT2VSampler is the text-to-video sampler for EasyAnimate's V3 and V4 generations, and it's the sibling to EasyAnimateI2VSampler in the same legacy lineup. Same idea: one node bundles prompt → video. Same catch: it's not for V5.1. The current models have their own EasyAnimateV5_T2VSampler, and that's the one you want for anything modern.

Why keep this around? Two reasons. First, V3/V4 checkpoints are substantially lighter than the 12B V5.1 monsters, so on a 16GB card these are genuinely more workable - the community ran them on 12GB cards. Second, V3/V4 do 144 frames at 24fps, so you get six-second clips where V5.1 tops out at 49 frames (~6s at 8fps but chunky motion). If long-ish, lower-quality video is your priority on limited hardware, the old line has a real argument.

The inputs that matter

• easyanimate_model - from LoadEasyAnimateModel, with V3 or V4 weights and the matching config entry (easyanimate_video_v3_... or easyanimate_video_v4_...). The default config is V5.1, so if you load a V4 model and forget to switch config, things will break in confusing ways.
• prompt / negative_prompt - STRING_PROMPT type, wired from the pack's text box node.
• width and height - defaults 1008×576, stepped in multiples of 16, up to 2048. Unlike the I2V node, T2V has no image to derive aspect ratio from, so you set the frame size directly. EasyAnimate is resolution-flexible by design; 1008×576 is the friendly starting point.
• video_length - 8–144, step 8, default 72.
• is_image - the fun switch. Set it to true and the sampler outputs a single frame - text-to-image, using the same model. It's a free static-image generator hiding in a video pack, and occasionally handy for previewing a prompt before committing to a 6-second render.
• seed, steps, cfg, scheduler - same set as the I2V sampler (Euler/Euler A/DPM++/PNDM/DDIM, default DDIM). No Flow scheduler here.

The output

images - an IMAGE batch of frames, ready for VideoHelperSuite to assemble into a file.

The honest take

Even at its peak this node lived in the shadow of the I2V version - the community's running joke about video T2V is that the model's prompt adherence isn't where the good stuff lives, and EasyAnimate's motion was never its strength. Use it when you have V3/V4 weights for the VRAM budget, or when you want 144-frame clips. If you're starting fresh, skip straight to the V5 sampler: better quality, flow matching, TeaCache for speed, and the model the pack defaults to.