Manual | Rendering options
Different ways of merging the images together to create the final stitch. Normally you want to keep it on Average
YUV images only. Upscales the chroma channels to match the resolution of luma channel, before merging them. Scaling takes quite a bit of time when done on a lot of images, so keep it off as much as possible. However when the stitch contains movement in both directions, the edges will contain artifacts, so you should enable it in this case. (#41 Make scaling alpha aware)
Some renders enable this silently, as they do not yet work without it.
Calculates the average of each pixels, and uses that as the output. Smoothest out compression artifacts and improves gradients, but because stitches usually require sub-pixel alignment, it can soften hard edges (lines) slightly.
Tries to take sub-pixel alignment into account by using non-linear interpolation. Very slow, and still tends to blur it slightly, making it only useful of experimental stuff.
Calculates the difference between all the values for each pixels. It is useful to determine if there is any misaligned images in the stitch, or if there are other issues affecting the Average render.
Use the Gamma option in the level tool to exaggerate the results, making it easier to spot potential issues.
Detects static content such as TV logos and credits, and creates a grayscale image, where black areas contain static content. Uses the algorithm explained here: spillerrec.dk - Logo detection and removal
Picks the darkest value for each pixels. This minimizes the effect of moving steam and rain in stitches, as it will pick the frame with the least likelihood of containing those. (See: spillerrec.dk - Too much steam in your anime?) The downside is that compression artifacts and the like is not removed.
Colors are currently handled incorrectly for YUV images: #3 Color incorrectly handled in Dehumidifier