Quality and workflow
Posted: Tue May 01, 2007 3:06 pm
Hi all
Im fairly new at the pano-game, but have managed to succesfully produce some fairly OK-ish panos. Id like to improve my quality versus size of my pano, sensibly, as opposed to playing around with all the different combinations of sizes, tile sizes, quality etc.
My workflow is something like this:
1) Shoot 6 fisheye (peleng 8mm+D50) images in RAW ( i dont bother with nadir and zenith, I adjust the max and min tilt to avoid the holes)
2) Covert to TIF(lossless) after finetuning exposure/whitebalance with Nikon batch editor.
3) I then stitch the images into a 2:1 equirectangular TIF image using Stitcher 5, using smart blending and maintaining 100% size and quality. Its a biig file!
4) Finetuning is then done in Photoshop CS2, adjusting the sharpness, curve etc.
5)I then save this image as a JPEG, 4000x2000, using a quality of 10(out of 12), still quite a big file.
6) This is then the image I import into Pano2QTVR, to produce the final QT tour.
7) I adjust the quality of the image using the quality slider, say 90%
8 ) I then adjust the tile-size to 200 pixels bigger than my width dimension on final window size, ie if its a 800x600 pano, I'll make the tile size 1000. This is the part Im unsure off.
9) I set the static quality to max, and the rotating quality to max.
My questions are:
1)At what point should you convert TIF to JPEG? If at all?
2)How do you determine the tile size to be used, as this dramatically changes the size. Its worth noting that I like to limit my FOV to a max and min value, this prevents the user from zooming into pixelated area, and also, avoids zooming out to the extent where it looks funny when rotating)
3)At what point is it best to sharpen and denoise the image? RAW or in CS?
4) What is the difference in "auto" and "optimal" tile size in pano2qtvr?
5) What size does the 2:1 image have to be for a) full size pano, b)800x600, c)400x300. Is there a method of determing the relationship. It seems like a waste of pixels to use a 4000x2000 image for a 400x300 movie?
If the users can share alternative workflows that produce good results, Im sure there's a lot of us that can benefit!
Thanx in advance!
gus
Im fairly new at the pano-game, but have managed to succesfully produce some fairly OK-ish panos. Id like to improve my quality versus size of my pano, sensibly, as opposed to playing around with all the different combinations of sizes, tile sizes, quality etc.
My workflow is something like this:
1) Shoot 6 fisheye (peleng 8mm+D50) images in RAW ( i dont bother with nadir and zenith, I adjust the max and min tilt to avoid the holes)
2) Covert to TIF(lossless) after finetuning exposure/whitebalance with Nikon batch editor.
3) I then stitch the images into a 2:1 equirectangular TIF image using Stitcher 5, using smart blending and maintaining 100% size and quality. Its a biig file!
4) Finetuning is then done in Photoshop CS2, adjusting the sharpness, curve etc.
5)I then save this image as a JPEG, 4000x2000, using a quality of 10(out of 12), still quite a big file.
6) This is then the image I import into Pano2QTVR, to produce the final QT tour.
7) I adjust the quality of the image using the quality slider, say 90%
8 ) I then adjust the tile-size to 200 pixels bigger than my width dimension on final window size, ie if its a 800x600 pano, I'll make the tile size 1000. This is the part Im unsure off.
9) I set the static quality to max, and the rotating quality to max.
My questions are:
1)At what point should you convert TIF to JPEG? If at all?
2)How do you determine the tile size to be used, as this dramatically changes the size. Its worth noting that I like to limit my FOV to a max and min value, this prevents the user from zooming into pixelated area, and also, avoids zooming out to the extent where it looks funny when rotating)
3)At what point is it best to sharpen and denoise the image? RAW or in CS?
4) What is the difference in "auto" and "optimal" tile size in pano2qtvr?
5) What size does the 2:1 image have to be for a) full size pano, b)800x600, c)400x300. Is there a method of determing the relationship. It seems like a waste of pixels to use a 4000x2000 image for a 400x300 movie?
If the users can share alternative workflows that produce good results, Im sure there's a lot of us that can benefit!
Thanx in advance!
gus