A clue to how Canon gets it’s progressive mode? MAYBE!

PappasArts

PappasArts

Veteran
Last edited:
Thank you so much for this. I have been telling people a few times on the dvinfo forum to take a look at the color from the 24F mode. What would be interesting to see is to make a chart with blocks of red green and blue at different angles. Then shoot the chart with a locked down camera in 30F and 60i. Also use the remote control to record some zoom in and outs of the chart to compare exact moments in time between the 30F and 60i. Using the remote gives a steady result with exact motion. I used this on my XL1 once for a poor mans motion controlled blue screen shot so I could match up different elements zooming in and out.
 
Your very welcome!

As an owner of both and over $16,000 invested in these two HD cams . I still will remain unbiased and could care less of fanboy antics. I was surprised my self when I saw this.

I was relieved to see that the HD100 does not do this.

I would rather Canon had taken the same 1.110 million progressive ccd's and did the formula that Panasonic used to get it's image for the HVX. I believe that would have rendered a better signal and true progressive image in the end. Someone should take this link to DVi in the XLH1 section where Steve Mullen has a good thread on this very subject.

Pappas
 
However all ccds use pixel shift. My guess is that JVC does too! JVC does not get affected by the filter. Canon is making a frame out of something, and it's not a de-interlace or flame blend. It would be easy to prove that if it did. What's obvious, is when this filter robs the prism from getting that spectrum to feed the ccd, it looses so much res that the image is very degraded. Who knows, well Canon knows!


Pappas
 
I don't see the mystery. My first guess was that it was the same "frame mode" as the gl2 and it still seems to be.

The JVC does not use pixel shift, nor does it need to. It records at native resolution.

Pixel shift moves the green CCD 1/2 pixel up (and sometimes to the left or right as well) and works like a bayer filter, kind of, to reconstruct extra resolution. The XL1h uses pixel shift, so blocking off the green sensor blocks off the source of its extra resolution above pure field blending. Hence, with the filter, it looks like field blending with no extra resolution.

Put a green filter over the hvx and you should get something closer to its 960X480 resolution, since you're blocking a whole ccd and all the resolution it adds. Try this test on the gl2 and you'll get similar results (although in SD, obviously.)
 
Very interesting Policar-- I was told pixel off set was used by all camera companies. I also thought Pixel shift and pixel off set was the same.

Barry Green should jump in on this one!

Pappas
 
Everybody uses it. The DVX is native-pixel res and it still uses pixel shift/spatial offset (same thing, two different names. Spatial Offset is probably the better name, because pixel shift makes it sound like it's actually shifting something, like a verb, whereas spatial offset more accurately describes it).

The spatial offset helps decrease fixed pattern noise and avoid stripes in the video that would come from having all the pixels aligned in a grid.

My guess is that Canon's using frame mode. Read up on Steve Mullen's "Progressive: What You Need To Know" article where he describes Frame Mode, and see if it matches what you encountered with the red filter...
 
Thanks Barry! See I thought I read long ago they all do it to prevent issues generated by ccd's.. It could have been Steve Mullen I read. That said, the Canon needs the green spectrum or it all goes to hell!


Pappas
 
This is silly. Why on earth would you want to completely deplete an entire primary color from the spectrum BEFORE it reaches a 3CCD block? These aren't B&W cameras. They're COLOR HD cameras.

How often would someone use this type of obscure color filter to recreate a B&W look anyway? Seriously? Wouldn't you want to recreate this look in post where you have FAR more control over the contrast range and black levels anyway?

It' what's coming out of the back of the camera that's important. It makes more sense to me to give the camera a fighting chance to process the entire signal, process that signal, and then re-treat the image after the fact. I'm sure if you turned off the entire green channel from the original color image you won't see this "anomally".
 
The purpose wasn't to show how much BW would suck oh the H1 with a red filter but to help show and figure out what might be going on to get to 24F. I understand it might look good but many of us wnat to know why and how it looks good.

The fact is that if a filter does this what happens to the resolution when you are shooting something with no green? For example shooting red baloons against a blue screen (I know, why the heck would somebody what to do that?). The lack of any green pixels might hurt the resolution of the image.
 
No. There's a big difference between letting the entire light spectrum enter the lens versus depleting the entire green portion of it. A blue sky with red baloon would still contain the entire spectrum of light to pass through. Therefore you'd have no issues with how the CCD's are processed. But filtering out 1/3rd of the spectrum BEFORE it hits the CCD's is entirely different.

Now, I agree that finding out how Canon's achieving their "look" is important. But I find the timing of this thread with the timing of selling Pappas' H1 suspect. It gives the illusion that something's wrong with the camera.
 
For the record, surely this isn't a RED filter if it is blocking green? We would want a magenta filter technically....
 
What color are the filters on each of the 3 CCDs? RGB? I don't think it's CMY.

And wouldn't a RED filter block both the blue and green spectrums only allowing the RED spectrum through. It's not surprising that the image looks like crap when you deprive the CCD block of 2/3rds of the available spectrum.
 
Wouldn't that be shooting myself in the foot when selling my own H1! I believe most people- not all would have sat on that info I found to serve there own needs then after selling would then bring it up. However I have much better character then that.....


Filming in B&W is another form of expression. Geez when were all are using D-Cams 15 years from now fully there will be no more B&W films I guess since all cameras will be color. For the record in B&W which I have been shooting since I was a young kid, has it's own contrast canvas to paint on. Filters like 25's - K2' etc help enhance that contrast or lower it. Shooting color for B&W is like shooting Color film for B&W it's a joke IMO, it shows a lack of commitment to the project and to doing it right for the medium. My JVC material work just like it does with D-SLR's or even the Kodak DCS 760m does when shooting B&W.

As for the 24f mode I could careless- It was last week when filming side by sides of the HD100 and XLH1 in B&W this was discovered. So I thought hey, maybe this is the clue Steve Mullen and party has been looking for!

My JVC B&W Footage contrast tests: http://homepage.mac.com/WebObjects/... BLACK & WHITE- &templatefn=FileSharing1.html



My JVC B&W no filter tests: http://homepage.mac.com/WebObjects/...F TEST FOOTAGE &templatefn=FileSharing1.html


starway2001@mac.com said:
But I find the timing of this thread with the timing of selling Pappas' H1 suspect. It gives the illusion that something's wrong with the camera.
Click to expand...


Pappas


Michael Pappas
Arrfilms@hotmail.com
PappasArts & Arrfilms Main site
CONTACT VIA AOL INSTANT MESSENGER
AT { PAPPASARTS2 }
XLH1 and HVX200 frame grabs and news here:
http://www.pbase.com/Arrfilms
http://www.PappasArts.com
http://www.Myspace.com/PappasArts
 
Last edited:
starway2001@mac.com said:
Now, I agree that finding out how Canon's achieving their "look" is important. But I find the timing of this thread with the timing of selling Pappas' H1 suspect. It gives the illusion that something's wrong with the camera.
Click to expand...

Starway, I can guarantee you that Michael would never have posted that pic if he had ulterior motives. Like he said, wouldn't it be shooting himself in the foot?

Mike's about as independent minded as they come, and I think it actually shows integrity and a brand agnosticism to post that pic. He's showing something of interest to the 24F puzzle, which I can appreciate because it is going out on a limb when he actually needs to sell the camera. (wish I could buy a second!!!!!!)

Starway, just like you I agree that for most purposes this isn't a typical use for the camera. Most will almost always shoot color but it is something to consider if intending to go the black and white experimental/purist route and shoot with a red filter. I find it interesting, but certainly not any kind of deal-breaker for the camera.
 
CCD's are inherently black and white sensor devices. So when you shoot with a 3 CCD camera, it's like simultaneously shooting three black and white images with red, green, and blue filters--the light usually being split into R, G, and B with a prism. So the end product is three channels that can be combined in any ratio to recreate the effect of shooting with any colored black and white filter (equivalent to using the channel mixer in Photoshop). If you add a 25 Red filter in front of the lens, you effectively remove the green and blue wavelengths before they can create the G and B channels. This is exactly the same effect as shooting with no filter and throwing away G and B channels in post processing. The black and white image is obtained by throwing away the chrominance data and keeping whatever luminance data you've collected.

The interesting part is that with any camera that uses pixel shift/spatial offset, you can't just discard the other channels without losing luminance resolution. I'd guess that if you say, recreated an orange filter (R=255, G=165, B=0) by mixing 65% of the R channel and 35% of the G channel, you'd see chroma-dependant luminance resolution decreases within the image. So depending on the color of an object within the frame, you'd see a different resolution. If you mix all channels equally and throw away the chrominance information, you'd have the full spatial offset luminance resolution.

Imagine if all three CCD's were layed end to end (or completely offset). You'd completely maximize the luminance resolution for a 300% increase at the expense of all of the chrominance data. So, spatial offset sensors are not the best solution if you want to be creative about a final black and white image, regardless of whether you use filtration (never capture some color channel information) of just discard color channel information in post-processing.



PappasArts said:
...Filming in B&W is another form of expression. Geez when were all are using D-Cams 15 years from now fully there will be no more B&W films I guess since all cameras will be color. For the record in B&W which I have been shooting since I was a young kid, has it's own contrast canvas to paint on. Filters like 25's - K2' etc help enhance that contrast or lower it. Shooting color for B&W is like shooting Color film for B&W it's a joke IMO, it shows a lack of commitment to the project and to doing it right for the medium.
Click to expand...
 
starway2001@mac.com said:
What color are the filters on each of the 3 CCDs? RGB? I don't think it's CMY.

And wouldn't a RED filter block both the blue and green spectrums only allowing the RED spectrum through. It's not surprising that the image looks like crap when you deprive the CCD block of 2/3rds of the available spectrum.
Click to expand...

At the same time however the exact same filter was used on the HD100 and it didn't have this issue. If it is clear that blocking the green and blue light would cause this issue then why does the HD100 look fine with the same filter blocking the green and blue light? The filter is blocking color not resolution.
 
You must log in or register to reply here.
Back
Top