site_archiver@lists.apple.com Delivered-To: pro-apps-dev@lists.apple.com User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.18) Gecko/20110617 Thunderbird/3.1.11 ARGH. Typos… Use this email instead! I think you're seeing a bug here. The easiest solution is to make your templates at 1920x1080. All of the templates we ship are at this size and work properly for smaller frame sizes. It appears that when you make a smaller template and attempt to use it at a larger size, some bugs show up. (Feel free to file bugs on this.) Okay I will try this. So I made a 1920x1080 template and applied it to a 1280x720 piece of footage in FCP. I get that the input is 1920x1080 and that the output image is 1280x720. So, in order to hand me the *1920x1080* image, do you pixel replicate? Scale using a bicubic? Etc.? What this test shows me is that I can't get the original pixels, which means that either (a) pixels will be replicated by FCP/Motion in order to give me the template size, in which case can you send the math about how you scale up so that I can pull the original pixels out of the bigger buffer; or (b) pixels will be scaled up using some filter (bicubic, bilinear, etc.) by Motion and my images will be unnecessarily soft upon output… because they will be scaled up with an kernel that doesn't necessarily preserve the input pixels anywhere in the*1920x1080* image you hand me. Can you describe what you do internally? Frankly, whatever you are doing seems like there is unnecessary processing going on at the very least (because you scale up and I have to scale down by throwing away pixels), and at the very worst, no matter what I do the returned images to the user will be softer than the original image, even if I just copy the input to the output. - Nathan _______________________________________________ Do not post admin requests to the list. They will be ignored. Pro-apps-dev mailing list (Pro-apps-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/pro-apps-dev/site_archiver%40lists.ap... On 07/26/2011 05:23 PM, Peter Litwinowicz wrote: We're seeing a similar issue here, but it seems to only affect images retrieved from input wells through the FxParameterRetrievalAPI -- the main input is coming out unscaled. Motion and FCP have always been braindead when it comes to well inputs, so we already have a bunch of code that rescales the images, fixes up fields, etc, and so far our plug-ins are handling this without complaint. But it's pretty nasty for the host to scale images up only for us to scale them back down (or vice versa). This email sent to site_archiver@lists.apple.com