|
|||||||||
|
Thread Tools | Search this Thread |
October 28th, 2003, 05:55 PM | #1 |
Major Player
Join Date: Jul 2003
Posts: 479
|
Got a PAL Xl1s...questions
In the manual it says that frame mode has 1.5x more vertical resolution, yet when I did resolution chart tests, interlaced mode certainly has more resolution. In frame mode, it could barely make 350 lines out..in 50i, it could do over 400.
Second, how would you guys suggest I do the image/pixel resize for an NTSC dvd? My goal is 24fps, so I've already taken some frame mode 25fps footage, slowed it down to 24 without resampling of course, and stretched the audio. It looks great on the computer, but how would I do the resolution adjustment to NTSC in after effects, premiere or avid. Also this is kinda related but here goes. My understanding of the 4:2:0(or 4:1:1, or any other) scheme, is that the camera when sampling the image from the CCD's records full luminosity, but skips some color samples. Regardless of the pattern in which it samples the color, doesn't the camera fill in the missing color sampled pixels with an approximation, such that when I load it up into my computer, it doesn't matter if it originated in 4:2:0 or 4:1:1? I mean, all the computer sees is a certain number of pixels per frame, and whatever color the camera assigned to them right? How would the computer know if it is 4:1:1 or 4:2:0? If it doesn't know then, I wouldn't loose any color information when I encode a DVD right?(excluding the resolution resampling) Thanks, Juan |
October 29th, 2003, 09:28 AM | #2 |
Major Player
Join Date: Jul 2003
Posts: 479
|
followup....
Ok here are the results of my first 24fps frame mode tests on the PAL XL1s, using After Effects.
1.Recorded some footage at 25p(frame) 2.Imported footage into After Effects, interpreted as progressive, and conform to frame rate 24fps. 3.In a NTSC 24fps composition, i did a transform and scaled the PAL image to fit the NTSC image. Since both images are 4:3, this works out almost perfectly if you click on the "include pixel aspect ratio' in the scale dialog. 4.Exported the file to an AVI uncompressed,and burned on a CD. I also included the original PAL 25p footage in the DVD. Basically, the results where REALLY good. The PAL footage looked identical to my eyes, with the exception that the top and bottom where cropped. The NTSC 24p adjusted footage looked great. I had a LOT of colors in the scene, including a flat red sweater, and the red actually looked good! I was very impressed, and I didn't notice any additional artifacts. The big question mark here is, that I am using a large TV as a monitor, but it looked just as good on the computer screen. The only tricky part seems to be getting the scaling just right. if it is a little off, the transformation gives some jagged edges. but if done right, the additional vertical resolution really shows up in the final detail. Now, i am still a little confused as far as the color space...is my DVD software then just doing encoding in 4:2:0? Because it is definitely putting out NTSC resolution, but the colors look fine. Or do DVD's simply work differently than digital tape formats? From what i've read, i'd expect to get 4:1:0 if I put PAL into NTSC...but on the other hand, as explained in my original post I don't understand why it would matter, once it has been loaded into a computer... Juan |
October 29th, 2003, 01:21 PM | #3 |
Trustee
Join Date: Oct 2003
Location: Toronto, Canada
Posts: 1,055
|
Although I am not quite there yet, soon I will be in the same boat as you as my XL1s PAL arrives tomorrow!
I have no intention of shooting in frame mode since too much vertical resolution is lost. To yield a FRAME mode look I will shoot 50i and deinterlace in post using built in deinterlace tools or DVFilm Maker software by www.dvfilm.com. To convert to NTSC I will plan on using Atlantis software also by www.dvfilm.com. This software does not change speed or video or audio, and gives the appearence of telecine transfer - which the DVX100 is most known and popular for. This would solve most of the issues you are talking about wouldn't it? Your way might be cheaper though! Dennis, |
October 29th, 2003, 02:51 PM | #4 |
Major Player
Join Date: Jul 2003
Posts: 479
|
Dennis,
You know, I do have DvFilm Maker and hadn't even tried it on 50i footage! I do have a DVX100 as well, and I have to say that I am getting very decent results with frame mode, slowing it to 24, resizing for NTSC and doing a small curve correction for the film look. By all means, the DVX100 is king in sharpness, but i'm going to try DVFilm Maker with the XL1se in 50i mode and see how much better that is than frame mode. My resolution tests showed a great deal more detail in interlaced mode. I thought the 4:2:0 issue would be a problem, but apparently it is not for DVD's. I am getting GREAT color, I think it might be up to par with the DVX. Although I have to admit the XL1s has quite a noisy output, even with the gain all the way down. One thing that has helped in frame mode is to turn the sharpness up a bit. I know, i know, this is a heresy in the film look book, but it just looks more detailed and i can't see any noticeable sharpening artifacts. Juan |
October 29th, 2003, 03:01 PM | #5 |
Major Player
Join Date: Jul 2003
Posts: 479
|
Good news...
So far, DVFilmMaker has given sharper results deinterlacing 50i footage than the raw frame mode footage. I setup the camera in front of a resolution chart, and in that exact same situation, recorded in frame mode and interlaced mode. The 50i deinterlaced footage resolves at least 100 lines more than the frame mode footage. Of course, there is still the problem that even and odd lines are recorded at different times, so there will be the combing effect on fast moving objects. I wish I could post pictures here. Juan |
October 29th, 2003, 03:13 PM | #6 |
Major Player
Join Date: Jul 2003
Posts: 479
|
I have posted framegrabs of the 50i and frame mode resolution graphs for the XL1SE at:
http://expert.cc.purdue.edu/~pertierr |
October 29th, 2003, 03:26 PM | #7 |
Trustee
Join Date: Oct 2003
Location: Toronto, Canada
Posts: 1,055
|
From what I understand about DV Film Maker, it analyzes the motion in each frame and only changes the pixels in that area instead of the whole frame like the deinterlacers built into Adobe Premiere, etc..
Having said this, running still footage of a resolution chart through DVfilm maker would not undergo much processing because really there is no motion. DVfilm maker only really "kicks" in when there are interlace artifacts in a frame. This is how it is able to give the look of FRAME mode but with a much higher resolution. If you use DVfilm maker on NTSC video then effectively you end up with 30P like footage, and with PAL video you end up with 25P like footage. Both are film like in nature, and both simulate a progressively scanned image. Although I am more of a Sony person I chose the XL1s PAL to use exclusively with the mini 35 system by P+S Technik. Thanks for the images! |
November 4th, 2003, 10:44 AM | #8 |
Trustee
Join Date: Oct 2003
Location: Toronto, Canada
Posts: 1,055
|
I've emailed Juan to try yet another little test.
A still grab from the DVX100 in 24P mode and a still grab from the XL1s in 50i regular mode run through DVfilm maker deinterlacer. You see, 50i run through a good deinterlacer is equal to the effect of FRAME mode without the loss in vertical resolution though. And FRAME mode or deinterlacing provides the motion feel of a progressive image. My bet is that the effective resolution will be the same of the two setups but also proves that the added pixel count on the DVX100 is needed more to produce adequate resolution in progressive mode, not much to do with "higher quality" which is what most are led to believe. If you want quality than you go up in CCD size not just pixel count. Hopefully Juan can try this out and post his findings! |
November 4th, 2003, 11:05 AM | #9 |
Major Player
Join Date: Jul 2003
Posts: 479
|
Will try it today when I get back from work.
As a matter of fact, i was doing resolution chart comparisons between the DVX100 in all modes and the XL1SE. Even in 50i, the XL1SE appeared to resolve quite a few less lines than the DVX in progressive mode. However, there was a problem which I didn't foresee, which is that the two view finders on these cameras do not show the entire picture and they both show different areas. To top that off, the wide angle lens on the DVX complicates things a bit. When I do this test, i'm going to hook them up via firewire to my computer and record directly to disk so I can see the complete picture and align the images correctly for a better comparison. I will also include de-interlacing results from After Effects and Premiere as well as DVFilmMaker. This last one doesn't seem to do a very good job if there is a lot of motion present... Juan |
November 5th, 2003, 10:24 PM | #10 |
Major Player
Join Date: Jul 2003
Posts: 479
|
Results
I've posted the resolution charts at
http://expert.cc.purdue.edu/~pertierr There's one for the XL1se(PAL) in 50i mode, unzoomed. The DVFilmMaker results where identical since there is no motion. The other files are for the DVX100(NTSC) in 24PA mode. One of them is with the res chart far away and zoomed in almost to 10x, and the other is not zoomed in. The zoomed in version has a bit less resolution in my opinion. The XL1se did pretty well, but the DVX100 is king of resolution in the DV world right now... :) let me know what you think... Juan |
| ||||||
|
|