Z Image using a x2 Sampler setup is the way
A viral workflow uses a dual-sampler setup to generate cinematic 2.39:1 aspect ratio shots for AI video.
A detailed workflow for generating consistent, high-quality source images for AI video has gone viral, shared by creator 'superstarbootlegs'. The method centers on the 'Z Image' model, praised for its speed and aesthetic, and employs a specific two-step sampler (t2i) setup. The creator starts by generating a small base image, typically at a cinematic 2.39:1 aspect ratio (e.g., 288x128 pixels), using a low denoise (1) but a higher CFG scale (4) to establish strong composition and structure.
This small image is then upscaled by a factor of six within the latent space and passed through a second sampler set to approximately 0.6 denoise. This second pass refines details and quality. The creator notes this configuration has proven most reliable after extensive testing. They have released updated ComfyUI workflow files ('Workflows 2') for others to download and use, streamlining the process of creating polished frames to feed into increasingly demanding AI video models like Sora or Runway.
In the same post, the creator also announced an update (v1.0.1) to their 'AIMMS' storyboard management tool, fixing popup issues and adding a right-click option to download media directly from the preview pane to speed up shot iteration. The post concludes with a critique of the 'Klein 9B' model, which the creator found to be slow and produce poor results, questioning if others have found a successful approach for it.
- The workflow uses a dual-sampler setup: a first pass at 288x128 resolution with CFG 4 for structure, followed by a 6x latent upscale and a second pass at ~0.6 denoise.
- Optimized for a cinematic 2.39:1 aspect ratio, the pipeline is designed to produce reliable, high-quality base images for feeding into AI video generation models.
- The creator has released downloadable ComfyUI workflow files and updated their 'AIMMS' storyboard tool to v1.0.1 with faster media download features.
Why It Matters
This provides a reproducible, high-quality method for creators to generate consistent visual assets, which is critical for coherent AI video production.