My conversion of Contingency to video moved ahead this week. I have about half of the images converted to video now. I had to overcome a few snags.
The whole film runs a bit over 3,800 frames, but since I already have video of some of the shots, I didn’t need to transfer all of them. It still left me with 2,300 frames to convert. This week I got 1,100 done, or about 48 per cent.
The week before I shot all the frames I wanted on the optical printer. It isn’t too difficult to import them into Premiere Pro and create a video. Unfortunately, I ran into a few problems.
The first was that the images were in RAW format, which I couldn’t import into Premiere Pro. I converted them to jpg’s with Picasa. I started with the same resolution as the raw files (3522 by 2348), but that really slowed down Premiere Pro to the point I couldn’t use it. I changed to 800 by 534, which is still larger than the 720 by 480 video I want to produce. That worked better.
The next snag was image jitter. Apparently the optical printer didn’t align each frame in exactly the same spot. I don’t know if it was a problem with the machine, or with how I loaded it.
Premiere Pro allows you to move the images around in the frame, so it was possible to adjust each frame to remove the jitter. That was easier said than done.
The first show wasn’t too bad because it was a title. I over lay a horizontal line and moved each frame so that the title aligned with the line. It was very finicky work though.
For the next shot, also a title, I overlay one of the frames on top of the other frames, switched it to a negative, which is called invert for some reason I don’t understand, then made it 50 per cent transparent. When I aligned the frame properly, I would get a grey screen. It never fit exactly, but it was easier and faster.
The rest of the shots were not tittles, which was a little harder. I used the same technique, but it turned out that since the camera moved during the shot, the edge of frame would move in and out of view. I had to go back and move every frame up a little so the edge wouldn’t show. I also had to zoom in, which took more time.
After a couple more shots I did a little research and found I could use the paste attributes feature of Premiere pro to save a bit of time.
Eventually I came up with another trick that made a big difference. I put the frame images into a sequence that set up as high definition. When I was done with the HD version, I would set up a standard definition sequence and put the HD version in the time line. Then I could move and zoom the whole shot to give the image I wanted.
That also allowed me to use the frame borders to align the frames. There were always imperfections in the border I could use for alignment.
I had a lot of trouble with Premiere Pro crashes. The process seems to have pushed the system to the limits. It got better as I got more practice.
I know that some video cameras have an image stabilizer function, which should mean that there is a software method to do all the realignment automatically. I wasn’t able to find one. I didn’t look too hard though.
The rest should take a couple of days to do. Then I need to incorporate all the shots into a final video. Some of the shots could use some colour correction. I hope to post the final video by the end of next week.