Large Still Images into PE - One Workflow

Question

Everyone wants the highest quality that they can obtain when doing their videos. It’s natural to want the best. Well, when dealing with still images, bigger is not necessarily better, for two reasons. First, overly large still images can really tax a system and second, one is limited to the frame size of the video, so these have to be resized somewhere - this resizing can be in the NLE (Non Linear Editor) program, or in an image processing program like PS (Photoshop), which does a better job anyway. Doing this in PS, or PSE, will result in better resized images, and they are easier for the NLE to work with. Quality is as high as your Project’s Preset will allow, and you are more efficient, with fewer crashes, slowdowns and hangs. It is a win-win situation.
Here is my normal workflow when dealing with still images. This workflow is for NTSC 4:3 720x480 with a PAR (Pixel Aspect Ratio) of 0.9. If your Project’s Presets are different, use those specs to resize to.
Since I shoot my still images in RAW, I Copy my files from the CF card to my system and catalog these images by location, subject and date (if necessary). I’ll do a quick conversion and Save_As Adobe DNG for backup. I then process these RAW images in PS with the ARC (Adobe Raw Converter), correcting them and then doing a Save_As PSD into a sub-folder. All of this is in my still photo library.
Normally, I will edit these PSD’s to find the images that I wish to use in a Video Project, and will Copy the selected images to another folder. You’ll see that I work with a lot of Copies, so my original files are always untouched and stored elsewhere. This guards against anything happening to them.
At this point, I’ll decide how I wish to use these selected images in my Video Project. Let’s just say that they are all horizontal images, and are still full-size from my camera. As stated, my Video Projects are DV-NTSC 4:3 720x480 PAR 0.9. [Remember, your Video Project may vary, so you will need to plug in the dimensions for YOUR Video Project in that case.] I also will have done my Cropping on each image individually, to get them to 4:3 Aspect Ratio. I do this my eye and by hand, rather than via an Action, because I want full aesthetic control.
In PS, I have a set of Actions for Video. An Action is like a Script, but less powerful and less involved in the writing. As I have already done all of my image enhancements and additional processing before I did my Copy to the selected folder, I only have to worry about my Action resizing these selected images for use in my Video Project. My Action here is to resize to 720x480 with a PAR of 0.9, and I normally use the Action that does this with a particular resizing algorithm, Bicubic-Smoother (though I also use Bicubic-Sharper on occasion).
For the next step, I go to my folder structure (remember, this folder contains copies of my selected still images in PSD format), and create a new sub-folder "[Project Name]_720x480." Back in PS, I choose File>Automate>Batch. Here I set my Source Folder, my Destination folder and the Action to perform. In my case, it’ll be the Destination Folder, that I just created, [Project Name]_720x480, and my Action will be my NTSC 4:3 720x480 Smooth. I check to have the Open command by-passed, because I do not need to see this take place on my monitor. When I hit OK, PS grabs all files in my Source Folder, runs the commands of my Action and does a Save_As for all files into my Destination Folder. I can process hundreds of large images down to a great 720x480 PAR 0.9 via Bicubic-Smoother interpolation, in moments. Now, I’m ready to go. Last, I Copy my Destination Folder to my Video Project’s folder hierarchy (usually on another HDD), and then Import these processed stills into my NLE.
What if I need to pan on one, or more of these images, while they are zoomed out completely? I don’t have enough pixels in my horizontal dimension to do this. I am just filling the frame with my still. Well, if I find that there are such images, I go back to my folder with the full sized images in my still images library, and select the ones that need to be larger. I run another Action on these, but it’s one that resizes to something larger than 720x480, say 1000x750. Now, I have another Destination Folder with the name [File Name]_1000x750. I’ll Copy this over to my Video Project, and Import these into the NLE. Here, I can go to Project Panel and remove the 720x480 versions if I so choose, but since a Premiere Project file (.PRPROJ or .PREL) is only an XML database, I may just leave them. It does not contain any media files, just links to where they are on the system and to what operations are performed on them.
By doing my resizing in PS, rather than in Premiere, I have accomplished two things:
1.) I have better quality resized images, using the algorithms in PS, plus have a choice of several interpolation methods to work with.
2.) I have lessened the processing load on my NLE and on my system, while doing the editing
I get higher quality and lower resource overhead - hence my reference to "win-win."
Now, back to my aesthetic control. I do not do any automatic zooming or panning. If one allows the NLE to do this, then they will want to probably process all of their images to 1000x750 (remember, this is for an NTSC 4:3 Project, so you will need to calculate what YOUR Project will require).
The two programs that I use are Photoshop and Premiere Pro, but Photoshop Elements can do the same things, though the exact commands might be different. Premiere Elements will handle the resized still images, just like Premiere Pro and the only difference will be the terminology used when one wishes to Import the still images.
I also keep all of my images in .PSD (the native format of PS), and do not convert to JPEG, or other. If one’s camera shoots only JPEG, I suggest writing the Action to do the Save_As to .PSD, as another JPEG compression will cost one quality. Yes, the JPEG’s will be smaller, but remember we are looking for the ultimate quality, so larger file sizes are just part of that equation.
One does not have to deal with all of the Copies, as I do. However, this allows me to go back to the originals, or to the processed full-sized .PSD’s at any step along the way. There is only one thing worse than not being able to go back to an intermediate version with full Layers and Adjustment Layers, plus any Alpha Channels, and that is finding out that you’ve lost your original RAW and DNG backups! That’s why I do a lot of Save_As and also work from Copies all along the way.
Hunt

Answer 1

Your workflow looks good. I do similar, but use PS, in lieu of LightRoom. I also do DNG's for my archives.
Provided that one chooses a JPEG compression algorithm setting that does not do too much compression, I doubt that anyone, but the most very critical, could tell the difference in Video. Most of my tests on PSD vs JPEG have been for print. There, one can more easily detect the differences. Video "hides" some of that.
To date, I have not had a Project where the Asset size differences between equally sized PSD's vs JPEG's caused any slowdown, or problem. There could be a resources savings with the smaller JPEG files, but there is a tiny bit of overhead dealing with the JPEG compression. I have never tested this, so can only guess that the smaller Asset size of the JPEG would trump that overhead - just a guess on my part.
For me, keeping the images in PSD does save a tiny bit of work in my Action (basically one less operation to perform), but I doubt that one could measure that time difference, even over the automation of hundreds of images. Besides, it's only one additional line in the Action. My feelings on JPEG vs PSD is firmly based in my print experience, and I am probably being too critical with images going to video. When I move up to HD and BD authoring, I need to apply a very critical eye, to see if I can tell the differences on an HD TV. So long as one does not apply too much JPEG compression, the differences should be very slight, at the worst, and maybe not even noticed, at best.
I do minimize the impact of many files on my Project by sizing to what I need. If I will not be doing any pans on zoomed-out images, I size to my Project. For pans on zoomed-out images, I calculate just what I will need for those pans, and might end up with several groups of sizes, to accommodate each. Still, the vast majority will be sized to exactly what I need for the Project - very few extra pixels.
In my case, and yours too, I have my RAW, my DNG, my working Layered PSD's, and then my sized output. I always keep all working PSD's, as I might change my mind, or my client might change theirs, and I do not want to have to go back and redo work, if I still have those working files. I also do as little destructive editing, as I can, using Dupe Layers, and Adjustment Layers, whenever possible. If I can, I never Flatten, or Merge Layers, so I can make any/all changes at any time, and only have to do the resizing via the same Actions. That is basically a "one-button" solution, once I have made the changes required.
Good luck,
Hunt