I was wondering if someone could explain to me what the benifits of Photo mode are. I can't count the number of times I've lost control of my camera because I accidentally hit the Photo button. It makes me wonder what good it really is. In Post, what is the difference between capturing a frame from the normal video stream and images shot in Photo mode? Another question, when you press the Photo button does the camera take the "photo" in the current mode (frame or interleave) or are full frames always recorded?

Ron,
Photo mode on the XL1s is really of no value. I suspect it was prompted by baseless marketing whims. It lays down 6 secs worth of one frame, which I think is non-interlaced (frame mode).

Photo mode on the new GL2 might be of some value since it actually operates more like a digital still camera. But even that's questionable.

Ken -

Thanks for the input. It would be interesting to know if "Photo" is always in frame mode regardless of how the camera is set. At least that would be a small redeaming feature. I shoot mainly in Interleave mode, but would like my stills in frame mode to avoid the potential time displacement between fields.

I forgot, but it messes with the format of the dv tape. I might be wrong, check all the specifics before you use that useless feature.

The only real use of photo I read of was to use it to mark segments on your video. You can take a picture of your slate, or whatever else will help you identify the segment when you are fast forwarding through.

If you shoot on interlaced-field mode, then hitting the photo button is a good way to have a deinterlaced single-frame to capture. If you shoot on frame movie mode, then the photo button is useless, because it's just equivalent to capturing a single frame from your frame movie mode video.

Robert -

Is the captured "photo" interpolated the same as frame mode video or does Canon slow things down enough where they can capture a true progressive frame? If it is interpolated, then your taking a resolution hit. I do mostly scientific work so shy away from Canon's "Doris Day" mode, except for the talking head shots that merit it.

I use photo mode when shooting artwork in galories or buildings. Also, It is great "B" role, with voice over. and adds just a touch of the "document" look to a production. I also use it a lot just to play with and check color. bob

Hi Ron,

I have no idea what the "official" Canon answer on this one is, but my tests with a Canon XL1S show no increase in high-frequency detail with photo mode versus a still-frame capture from frame movie mode video. Further, if you take a difference image between the two captures, you'll end up with a black frame (as expected). Now, the histogram of said difference image shows that the difference image isn't completely zeroed, i.e., there are pixels with nonzero values. However, the this histogram looks similar to the histograms of both difference images between two photo mode images of the same still frame (taken with the remote control, so as to prevent movement of the camera when recording and pausing) and between two still captures taken from video shot in frame movie mode. One can conclude that the only subtle differences between the two images are due to compression artifacts inherent to DV encoding.

So, my answer is, the photo mode employs the same interpolation algorithm as the frame movie mode in the Canon XL1S.

If you'd like to see the two images I used for comparison, I've uploaded them to my server here. Caution! These BMP format files are about a megabyte each!

http://www.robertks.com/xl1/framemoviemode.bmp
http://www.robertks.com/xl1/photomode.bmp

These probably won't last forever, so those of you reading this post in the far future might not see these images. (Sorry!)

I'm afraid I don't get the Doris Day reference, so I hope you'll clue me in.

I'm afraid the "Doris Day" reference really dates me. Back in the 50's and 60's there was a singer / movie star named Doris Day. As they years caught up with her the director would shoot her closeups with a silk stocking stretched over the lens. As the years passed they apparently added more layers. By the time she retired, they were diffusing the image so much you could barely make out her features.

Thanks for the test shots. You should add one more to the series - a frame captured from video footage shot in interleave mode.

Interesting to know about Doris Day. Incidentally, I had a professor at USC (Drew Casper) who is the World's Foremost Doris Day Expert. If there was an AMC documentary on Doris Day, he'd be the scholarly commentator. No matter what course he was teaching--Kubrick, Hitchcock, Lucas--he would find a way to tie the final lecture in with Doris Day somehow, so that he could burst out into a Doris Day musical number for his finale, always to a standing ovation. Entertaining guy. Strange, yes, but que sera sera.

I just ran some tests to see if I could learn any more about Photo mode. I shot a still image as interleve video, photo while in interleave, frame video, and photo while in frame mode. In Premiere I captured still frames from each. The last three looked virtually identicle (apparently Photo always uses frame mode), and all three looked better than the first (interleave video).

I use ScenalyzerLive as my capture program and was suprised to see that it flagged "Photo" mode footage differently than video (frame or interleave). This is a good thing in that I usually use a DTR and not the camera for inputting to the computer. Index marks are not detected with this arrangement, but since Photos are, I'll use it in the future to mark my scenes. I knew it had to be good for something.

Very interesting. Can you explain the usefulness of Photo Mode with Scenealyzer a bit more for those of us that haven't used Scenealyzer? What does Scenealyzer do with Photo Mode video differently that makes it useful for marking scenes? Thanks...

Scenalyzer is a video capture program. It's rather unique in that it scans the tape at FF speeds. A 60 minute tape can be analyzed in 5 minutes. Normal scene detection is done by looking for jumps in the timecode. Basically it scans the tape and provides a map of, and a few frames from each scene it detected. You can then select which ones you want to capture and it locates them and transfers them to disk. One problem with this approach is separating the takes from the scenes. To the software, they are one of the same. Now when I have a scene in the can, I can lay done a "Photo" as a marker before moving on to the next scene. Perhaps a small advantage, but it will save time since I don't often shoot a production as a linear sequence.

Robert--

What did you use to capture those images you posted?

Although I never use my camcorder for dig photography, I recently explored a number of "photomode" concepts on camcorders. Conclusion: the cheapest digital photocamera beats even the best camcorders. Simply because the CCD structures, number of pixels, shuttering and readout architectures are not optimized for digital still photography. Several "concepts" (tricks) are used and "sold" by smart marketing people. The main problem is the basic interlace CCD readout and memory structures in a consumer camcorder ( and the professional progressive ones don't care about "the other application") The "solutions"go from simple line doubling through line interpolation, over the RB (even) and G(odd) concepts from Canon and more recently the use of a mechanical shutter for delayed readout of the interlaced CCD structure (Sony PC110, GL2(?)). Although the shutter concepts doesn't show (shouldn't) resolution loss the total number of pixels is still far below the cheapest dig camera..