Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Dsmc..

Status
Not open for further replies.
DSPographer said:
DPreview has posted two new video clips from the 5DII. In the second one here:
http://s3.amazonaws.com/movies.dpreview.com/CanonEOS5DMarkII/Video2.MOV
there looks to be strong color aliasing artifacts along the bottom of the train and on a grate in the upper left of the frame near the end. It could just be that the optical anti-aliasing filter is too weak but it is so visible that I think it may be caused by readout sub-sampling of the sensor in video mode. It looks like the bar is not yet set so high for a video mode of Red's DSMC.
Click to expand...

That video is quite unimpressive. Which is bad news for me, because I would LOVE to be able to shoot really remote 1080p video with the 5Dm2. :bye2:
 
Im sure Red will have its DSMC be great for motion as well as still.:wink:
 
Tom said:
That video is quite unimpressive. Which is bad news for me, because I would LOVE to be able to shoot really remote 1080p video with the 5Dm2. :bye2:
Click to expand...

Indeed so, the real show killer is the aliasing due to undersampling. OK, I'm healed and I will wait for news here :calm:
 
What do you mean by undersampling? Are they basically skipping pixels? How does their system even work?
 
Tom said:
What do you mean by undersampling? Are they basically skipping pixels? How does their system even work?
Click to expand...

Given the aliasing effects and a bit of arithmetics, it is quite obvious how it works. Their sensor would have roughly 3 times the columns and much more than 3 times the rows for 1080p. If you want a relatively fast rolling shutter, the easiest is to read out only every third line. To read out a line faster, you probably will read out only every third pixel.

Nikon does the same, look here for the resulting artefacts :

http://www.vimeo.com/1816867


Red chose differently, when shooting in a different resolution to be faster, they crop. That may be annoying from a certain perspective, but the result is better.
 
The difference is mainly that they both skip reading lines (a few of them) for 1080P because the sensor is too slow. We read every line for every format. Skipping lines is... not good.

Jim
 
I hope the autofocus is getting as much attention as image quality. It was frustrating losing shots with my Canon 1D Mark III that my old Mark II had no problem getting.
 
Hi Jim and RED,

I have 2 suggestions for RED to look into for your up coming cameras.

It would be Amazing if you offered the option to have 3DV Systems DeepC chipset in your cameras. This would allow the user to shoot footage without a green screen but still be able to key the footage based on depth. I can see many uses for depth channel information to be used in post.

http://gizmodo.com/gadgets/zcam-dep...and-minority-report-without-gloves-334426.php

I was wondering if you guys had heard about black silicon?

"Black silicon offers up to 500 times more light sensitivity than normal silicon"

http://www.dailytech.com/article.aspx?newsid=13188

I hope to see RED using these two technologies in the near future :)

Ben
 
To me the Canon D5MKII footage looked like 720p upsampled to 1080p.
What both Nikon and Canon are doing is just saving the Live View. On the Nikon you even have to hold a button while filming to avoid a stupid robot constantly moving the ISO while you roll. I'm still going for a Nikon till you guys are done with the DSMC. :) I rather have 4:4:4 sorta there than a Canon 4:2:2 (not sure about the lat 2, could be 1).

Should I have repeated something from an earlier message... It's 40 pages!
 
h.264 is generally 4:2:0

Graeme
 
4:2:2 came about due to the interlaced nature of video. Doing 4:2:0 on interlace is very tricky, and probably would not have been possible with the old analogue ways of doing things.

However, for progressive, if you think 4:2:2 makes sense, in that generally you can't see the difference on broadcast with the half rez chroma horizontally, well you probably can't see half rez chroma vertically either. 4:2:0 makes total sense for progressive video. With 4:2:2 you're saying vertical chroma resolution is more important than horizontal chroma resolution, but I don't think that is really the case.

As for the HDMI output - it all depends on the downsampling and internal processing as to if it's "worth it" in terms of chroma sampling. It may not be.

Graeme
 
i wanna see those test of what comes out of the hdmi port on the canon , it probably is less compression. The tests done with the hv20 sure looked different when the information was pulled directly from the hdmi - then again - i had an hv20 for awhile and there was just no affordable, portable and convenient recording option available for the hdmi port - this seems to be changing - is the cineform recorder a reality yet
 
Graeme Nattress said:
4:2:2 came about due to the interlaced nature of video. Doing 4:2:0 on interlace is very tricky, and probably would not have been possible with the old analogue ways of doing things.

However, for progressive, if you think 4:2:2 makes sense, in that generally you can't see the difference on broadcast with the half rez chroma horizontally, well you probably can't see half rez chroma vertically either. 4:2:0 makes total sense for progressive video. With 4:2:2 you're saying vertical chroma resolution is more important than horizontal chroma resolution, but I don't think that is really the case.

As for the HDMI output - it all depends on the downsampling and internal processing as to if it's "worth it" in terms of chroma sampling. It may not be.

Graeme
Click to expand...

Yes, but color space is only part of the equation, right? The video that the 5Dm2 records to CF cards is still subject to the H.264 compression downgrading, whereas if you could record the HDMI signal to Cineform, for example, the picture might look better. As you say, no one will know until the camera hits the streets, really.
 
Lack of compression might just show up more of the shortcuts in the downsampling of the whole sensor... I certainly see some vicious aliassing/moire in some shots...
 
maybe i can bring the 5Dm2 down to cineform and do a little test once i get my hands on it, assuming the cineform guys don't get the 5Dm2 first.

that's gonna be the only way to know.
 
Why is anyone here worried about D90 or 5D MarkII? Red doesnt need to even bothered talking about it. I can understand maybe, Ikonoscope A-cam DII but there isnt ANY camera out there yet that can compete with Red in any camera range(price/tech.)?:red_bandana:
 
Ayoji said:
Why is anyone here worried about D90 or 5D MarkII? Red doesnt need to even bothered talking about it. I can understand maybe, Ikonoscope A-cam DII but there isnt ANY camera out there yet that can compete with Red in any camera range(price/tech.)?:red_bandana:
Click to expand...


if their isnt any other camera that can compete with red in any catagory.... then why would they need to even talk about it at all... sure its there.... but obviously not even remotly in the league's they are going to produce ....

just my 2 cents
cheers
chris gus
 
Because by next year those two cameras might be greatly improved and truly compete. The companies behind them have plenty of money in their pockets and some of the best tech people in the busines working for them. I am absolutely sure they won't be sitting around doing nothing about making these cameras much better.
 
Status
Not open for further replies.
Back
Top