Differences between the Canon progressive modes

Canon's 24f and 30f modes produce amazing images, hard to distinguish from true progressive scan. Since it's their closely guarded secret, we can only speculate on technicalities, but does it really matter what kind of internal processing is done inside XL-H1 and new XH line if the final quality is so close to $20k XDCAM HD?

Canon's F mode (not to confuse with Frame mode of their first XL camcorders) achieves more than 800 lines of effective resolution and at the same time images have pleasant filmic appearance. What more can we expect at this price level?

Canon's 24f, 25f and 30f images are just too good to be simply deinterlaced. The best software deinterlacers do not achieve that quality (let alone the speed) simply because it's really not easy to deal with time displacement between the fields. It does not matter if we go from 60i to 24p or 30p or from 50i to 25p. The problems are there, sometimes pretty bad, especially when we deal with fine motion of high contrast details in the image. All of that makes me think Canon phase shift signal clocking the CCDs, scanning both fields at the same time. Since the imagers are technically interlaced and the image is not captured in one sweep but with two parallel scans, the process can not be called true progressive. OK. Resolution drop in f mode vs interlaced? I think it's the result of necessary filtering applied to both scans, but if the final quality is better than anything on the market below $20k, who can seriously complain here? :)

I doubt anyone watching our movies will notice the difference between 24f and 24p, but surely people will notice the difference in resolution and overall image quality. The best resolution CCDs in the class (1440x1080) and the results Canon achieves at this price level are truly impressive.

Cheers.
 
Last edited:
Excellent first post, and welcome to the forum! Canon of course won't say, but I have heard vaguely of this "phase shift signal clocking" technique, but I have no clue as to how it's done.

I think youre right about the deinterlacing thing. I've done all kinds of renders with the best algorithms and "smart deinterlacing" was never 100% perfect. By contrast, I have never seen a tell-tale deinterlace artifact from the HDV F modes.

800 lines of effective resolution? Only with 1080i, but I hate the whole resolution numbers game anyway. I've edited a fair amount of very good 24p CineAlta footage in HD, and I can say without any hesitation that the XL-H1 comes the closest to this standard.

It's not perfect, but you're right. Potentially now at an even lower price level than the H1, how can anyone complain? It's a dream for no-budget indies.
 
MovieSwede said:
Its not the F mode that bothers me

its the HDV.
Click to expand...
Buy the G1 and wait for the external HD-SDI recorder portable device option and the 10-bit Cineform RAW.
 
Elton said:
Do you think HDV shooting means you have to EDIT in the format?
Click to expand...
Good effort Barlow. There is a lot of misunderstanding here on the subject. I didn't want to mean about MovieSwede. He knows what he's talking about. But there is who insists on that. HDV is a format like any other. With its limitations. A non-broadcasting HD format, unless as 15% or 25% of the total footage if delivered to Discovery HD channel. But with Canon you have the HD-SDI or even the uncompressed HD port. Analog in the latter case, OK but there are more than one option around the corner. I prefer it rather than the lower res HVX and its lowlight response, for example.
 
I also heard a theory about phase shifting (or inverted cycle) of one of the CCDs (possibly G), but that would still result in time displacement, while Canon's images do not display any artifacts suggesting use of this technique. It's interesting. Again, it makes me think they use a kind of solution that falls in between interlaced and progressive acquisition, gathering images in two parallel sweeps, but requiring additional filtering. Anyway, I'm happy we have that quality available for big chunk less than $10k.

Some tests and camcorder comparisons available on the Web were not conducted properly. I dare to say some were quite biased against Canon XL-H1. When 24f or 30f footage is imported properly (e.g. with Premiere Pro and Aspect HD), I can clearly see vertical resolution of 800 lines, but what matters more is it behaves very nicely in fine motion. In currently commercially available technology, only Cinealta provides better results.

I am convinced Canon's HDV compression in F mode uses progressive scans which is much more efficient than interlaced compression (and that's why it can not be played back on other HDV equipment). At 25mbps rate, it produces really impressive results with minimal motion artifacts. For many of us it's a dream come true, even without HD SDI option.
 
Pixel shift is standard on most cameras today, including H1 and it works in both interlaced and progressive modes. I don't think it's related to f mode specifically. Shifting CCD horizontally or in both directions will not eliminate problems associated with interlaced scan. Imho Canon did not go this way. Seems way too hard to produce such nice progressive look real time in the camera, even with advanced processor.
 
As a matter of fact, the XL-H1 pixel shifts horizontally to achieve full-raster SDI spec. The CCD's pixels have a wide aspect ratio of 1.212...if you multiply 1440 x 1.212 you get 1745 horizontal res, which is then pixel shifted to achieve 1920.
 
Bogdan said:
I am convinced Canon's HDV compression in F mode uses progressive scans which is much more efficient than interlaced compression (and that's why it can not be played back on other HDV equipment). At 25mbps rate, it produces really impressive results with minimal motion artifacts. For many of us it's a dream come true, even without HD SDI option.
Click to expand...

You're correct. The camera's MPEG2 encoder is fed a progressive image in F mode, which generally compresses better and is more artifact resistant than regular 1080i mode. I also believe it has a shorter GOP cycle in 24F, but I'm not absolutely certain.
 
Elton said:
You're correct. The camera's MPEG2 encoder is fed a progressive image in F mode, which generally compresses better and is more artifact resistant than regular 1080i mode. I also believe it has a shorter GOP cycle in 24F, but I'm not absolutely certain.
Click to expand...

Shorter GOP in 24F makes perfect sense with extra 20% bandwidth available compared to 30F.
 
Emanuel said:
A non-broadcasting HD format, unless as 15% or 25% of the total footage if delivered to Discovery HD channel.
Click to expand...

Wanna know something funny? I just recorded a Discovery HD program last night called "Iconoclasts" which was originally produced for the Sundance Channel.

I swear it was all Sony HDV. I know the look well, and it was unmistakeably Z1u. It looked like it was shot CineFrame 30. A tiny bit soft for HD, but still decent...although you could definitely see a difference between the preceding program shot mostly on Varicam and CineAlta (it was called "Discovery Atlas") and the Z1.

You know what though? The show was about Robert Redford , Paul Newman, and their history together...and guess what? Content trumped technical specs, and it probably didn't hurt that the show featured lots of footage of both famous actors hanging out and waxing philosophical about their movie experiences.

Discovery HD will apparently bend the rules for acquisition as long as it features famous people or the content is very compelling.
 
You must log in or register to reply here.
Back
Top