SeedVR2 Tile Splitter (Upscale Factor):
The SeedVR2TileSplitterByFactor node is designed to facilitate the process of splitting an image into smaller, manageable tiles based on a specified upscale factor. This node is particularly useful when you want to increase the dimensions of an image by a certain multiplier, such as doubling its size. By allowing you to specify the desired upscale as a simple multiplier, this node automatically computes the necessary tile upscale megapixels, ensuring that each tile is processed efficiently. This approach not only helps in managing VRAM usage but also ensures that the final output maintains high quality by seamlessly integrating the upscaled tiles. The node is ideal for scenarios where you need to upscale images while maintaining control over the tile size and overlap, thus providing a balance between performance and output quality.
SeedVR2 Tile Splitter (Upscale Factor) Input Parameters:
image
The image parameter is the input image that you wish to split into tiles. It is expected to be in a format that the node can process, typically a single image rather than a batch. This parameter is crucial as it serves as the base for all subsequent operations performed by the node.
tile_size_mp
The tile_size_mp parameter specifies the maximum size of each tile in megapixels. It ranges from 0.1 to 16.0, with a default value of 1.0. This parameter is important because it directly affects the amount of VRAM used per pass; smaller values mean less VRAM usage, which can be beneficial for systems with limited resources.
upscale_factor
The upscale_factor parameter determines the multiplier by which the image dimensions will be increased. It ranges from 1.0 to 16.0, with a default value of 2.0. This parameter is essential for defining how much larger the final output will be compared to the original image, with 2.0 meaning the width and height will be doubled.
overlap_percent
The overlap_percent parameter defines the percentage of overlap between adjacent tiles. It ranges from 0.0 to 40.0, with a default value of 10.0. This overlap is crucial for ensuring that there are no visible seams between tiles in the final stitched image. Increasing this value can help mitigate seam issues but may increase processing time.
feather_blend
The feather_blend parameter controls the blending of the overlap zone between tiles. It ranges from 0.0 to 1.0, with a default value of 1.0. A value of 1.0 results in a smooth linear feather, while 0.0 results in a hard cut. This parameter is important for achieving a seamless transition between tiles in the final image.
SeedVR2 Tile Splitter (Upscale Factor) Output Parameters:
tiles_tensor
The tiles_tensor output contains the individual tiles extracted from the original image. Each tile is processed according to the specified parameters, ensuring that they are ready for further processing or upscaling.
meta
The meta output provides metadata about the tiling process, including information about the grid configuration, tile dimensions, and overlap settings. This metadata is useful for understanding how the image was split and for any subsequent operations that may need to reference this information.
resolution
The resolution output gives a hint about the resolution of the SeedVR2 process, typically calculated as twice the tile_size_mp. This information is valuable for ensuring that the upscaling process maintains the desired quality and resolution.
SeedVR2 Tile Splitter (Upscale Factor) Usage Tips:
- To achieve optimal results, start with the default
upscale_factorof 2.0 and adjust based on your specific needs. This provides a good balance between quality and performance. - If you notice seams between tiles in the final image, consider increasing the
overlap_percentto 15-20% to improve blending. - Use a
feather_blendvalue of 1.0 for smooth transitions between tiles, especially when working with images that have complex textures or patterns.
SeedVR2 Tile Splitter (Upscale Factor) Common Errors and Solutions:
"Tile Splitter expects a single image (batch=1), got {image.shape[0]}."
- Explanation: This error occurs when the input image is not a single image but a batch of images.
- Solution: Ensure that the input to the node is a single image. If you have a batch, extract the individual image you want to process.
"ValueError: Invalid upscale factor."
- Explanation: This error may occur if the
upscale_factoris set outside the allowed range. - Solution: Check that the
upscale_factoris within the range of 1.0 to 16.0 and adjust it accordingly.
