Wan 2.2 Cinematic I2V + InfiniteTalk - 24GB-Tuned ComfyUI Workflow Pack (720p) Devlog - itch.iohttps://adel93.itch.io/wan22-comfyui-workflow-packThree tested Wan 2.2 ComfyUI workflows for one 24GB GPU: 720p I2V (8-step + 4-step Lightning) and a talking head.https://adel93.itch.io/wan22-comfyui-workflow-pack/devlog/1668977/infinitetalk-on-a-24gb-gpu-the-window-setting-that-silently-never-finishesInfiniteTalk on a 24GB GPU: the window setting that silently never finisheshttps://adel93.itch.io/wan22-comfyui-workflow-pack/devlog/1668977/infinitetalk-on-a-24gb-gpu-the-window-setting-that-silently-never-finishesFri, 18 Sep 2026 22:59:57 GMTdevlogIf you are trying to run InfiniteTalk on a 24GB card and the render never finishes, the setting you want is frame_window_size - and the failure mode is silent, so it is worth knowing about before you...

]]>https://adel93.itch.io/wan22-comfyui-workflow-pack/devlog/1668958/wan-22-on-a-24gb-laptop-the-4-step-lora-was-slower-and-the-widget-bug-that-would-have-shipped-silentlyWan 2.2 on a 24GB laptop: the 4-step LoRA was slower, and the widget bug that would have shipped silentlyhttps://adel93.itch.io/wan22-comfyui-workflow-pack/devlog/1668958/wan-22-on-a-24gb-laptop-the-4-step-lora-was-slower-and-the-widget-bug-that-would-have-shipped-silentlyFri, 18 Sep 2026 22:27:14 GMTdevlogI spent the last stretch getting Wan 2.2 image-to-video to run well on a single 24GB laptop GPU, and three of the resulting workflows are now packaged up. This post is the technical part - the numbers...

]]>