Video Enhance - Load Anchors (Meta Batch)

In the Video Enhance pipeline, the anchors are the frames that actually get upgraded: a spaced-out set of full frames from your video, each one run through Z-Image to become a high-quality guide that LTX then uses to enhance every other frame. But here's the practical problem - you can't just feed all of them to Z-Image at once. You need to feed them one managed batch at a time, and that's exactly what VRGDG Video Enhance - Load Anchors (Meta Batch) does.

It's the bridge between the Prepare node and the Z-Image enhancement stage, built specifically to work through the VHS Meta Batch Manager (from the Video Helper Suite). VHS Meta Batch is the community-standard tool for processing frame sequences in controlled chunks, and this node plugs the anchor loading into it so Z-Image receives the managed sequence in one queued workflow instead of you babysitting a loop.

The two inputs

• video_enhance_context - connect Prepare Video and Anchors. The tooltip explains what it contains: the automatically generated full-frame anchor folder. This is where the anchor frames come from.
• meta_batch - connect a VHS Meta Batch Manager, and set its frames_per_batch from Prepare's anchor_count output. The tooltip is direct about the wiring: set frames_per_batch from Prepare's anchor_count so Z-Image receives the managed anchor sequence in one queued workflow. Get that number right and the batch loop covers every anchor exactly once.

What it outputs

• anchor_images - the current batch of anchor frames as an IMAGE tensor, ready for the Z-Image enhancement (or whatever node processes them).
• mask - a mask output, present so downstream mask-aware nodes (inpainting-style composition, if you use it) have something to work with.
• batch_frame_count - how many frames are in this batch; handy for sanity-checking that the meta batch is sized right.
• video_enhance_context - the context, passed through to the Store Enhanced Anchors node that follows.

Why the meta batch matters

The whole enhance workflow is designed to respect VRAM and queue discipline. Z-Image is small enough to run on consumer cards, but processing an entire video's worth of anchors as one giant batch will blow past what fits and make the queue unruly. Batching through VHS Meta Batch keeps each queued step a known size, which is also what lets the whole anchor pass run as one continuous workflow rather than you manually kicking off dozens of renders. And the deterministic order matters: Store Enhanced Anchors later depends on anchors arriving in a fixed sequence so the enhanced results map back to the right frames.

Install

Standard pack install (Manager → search vrgamedev, restart, hard-refresh). This node is meaningless without the Video Helper Suite (VHS) installed - that's where Meta Batch Manager comes from, and the pack's Video Enhance workflows expect it. If you load a workflow and VHS nodes show as missing, install VHS first, then re-run. The anchor enhancement side also expects a Z-Image checkpoint (the pack's Z-Image upscale/refinement workflows show how the author wires it), so have Z-Image Turbo or Base available in your model folder before you set up this stage.