Did you happen to follow the thread in image-SIG about how to make Photo use the new Antialias filter in PIL 1.1.3? I know you prefer Imagemagik, but for us poor windoze folks that don't have a unix emulator, we are stuck with PIL. Anyway, Fredrick suggested adding the Antialias filter to the resize function in photo.py. I did and it works great. All of the generated images look 200% better!

Great product! I would love to see Photo work with ExternalFile's External Folder to batch read the source images on the file system and leave the image original (download version) at that location and have the option of sending the generated thumbnail there and all the other variations to the ExtFile folder - perhaps a checkbox next to the display sizes that indicates to save in the ExtFile folder under var or to save wherever the source was found (zodb or External folder).

Part of why I mention this is that the contents of ExtFile's folder could be cleared like a cache, no risk of deleting a source image, no need to back it up, and if a display image is missing it can be regenerated on the fly. This helps a migration to a new server as the ExtFile folder can be ignored and the source images remain in their original file system folder context. Whole trees of images can be put into the file system and then External Folder can batch traverse and Photo can intercept each Image that External Folder is adding and add it as a photo instead as it does now. I think its just a change in the object filter settings to trigger the process on either form of added image.

I really think this is worth a look because of the power of the image gallery that can be created and that maintainability is improved as whole branches of the directory tree are pruned and relocated. Simply refresh the metadata view of the file system via a traverse script that runs the external folder batch script and the new structure and all the displays are regenerated using settings acquired from the photo folder that is in the zodb portion of the path.

On another subject, have you given any thought to pulling in the additional data fields in a jpeg typically written by digital cameras or captions and keyworks provided by image viewers as added properties of an imported photo? If not, are you aware of anyone having an enhancement or plugin for this? It seems like a straight forward loop through the properties that can be done when a photo is being added that has this extra data from the digital camera. I know I can spend time deconstructing the code to figure out where and how to add the attributes as each display onject is generated. I'm just wondering if you might already be thinking about this and know where the optimum insertion points might be to simply add all attributes found in the source file to each generated photo object. Please let me know your thoughts and suggestions.