Batch Execution

In the current version, you can click Run on multiple nodes in the canvas and let several tasks enter the execution flow. Different workflow types behave differently during scheduling, so plan the number of tasks according to your local ComfyUI performance and external API account capabilities.

Current Execution Rules

For local ComfyUI workflows, tasks are processed according to the ComfyUI queue. If the first local task is running and you click Run on a second local task, the second task will wait and usually starts after the first task finishes.

This means local tasks can be submitted from multiple canvas nodes, but actual execution speed still depends on ComfyUI, GPU performance, VRAM capacity, and the current queue state. For production projects, test one task first, observe runtime and VRAM usage, and then decide whether to submit multiple local tasks in sequence.

For external API workflows, concurrency depends on your API service account, plan, and platform limits. Some API platforms support multiple concurrent tasks, while others limit concurrency or process tasks in a queue. Before submitting multiple API tasks, confirm whether your account supports concurrency and whether concurrent calls may incur extra cost or trigger rate limits.

Usage Tips

• For local ComfyUI workflows, run a small test before submitting multiple node tasks.
• Long video, high-resolution, or multi-model tasks use more VRAM and time, so avoid submitting too many at once.
• For API workflows, pay attention to account quota, concurrency limits, and platform billing rules.
• If a task fails, check the error message on the node and review the model setup, input assets, or API configuration.

Future Batch Features

Future DFCine versions will continue to improve batch execution. Planned capabilities include more flexible multi-account and multi-GPU assignment, so teams can distribute tasks according to different workflows, models, and hardware resources.

Please follow DIGITFOLD version updates and official announcements for the latest progress on batch execution.