LINKS TO XL H1 CLIPS
Posted 26 January 2006 - 03:40 PM
Here are the clips:
http://www.hdvinfo.n...keseq24.m2t.zip (136M) 30f, incorrectly labeled 24
http://www.hdvinfo.n...panning.m2t.zip (43.6M) Here's a 24f clip
http://www.hdvinfo.n...ssingby.m2t.zip (20.2M) Another 24f clip
You will need HDTVtoMPEG2 to rewrite the files as MPEGs so you can open and look at them. The conversion is lossless.
And now, my thoughts:
First of all, after having actually SEEN THE FOOTAGE from the XL H1, I can say it is absolutely, positively 60i in all forms, even 24f. David was correct in the other topic when he predicted that the 24f video is written as 60i with 3:2 pulldown. There may be a mode to write it without pulldown, but I have not seen any footage that way, only 60i with 3:2 pulldown. I doubt it, however, as this would not conform to the HDV spec and would not be readable by any other post gear except another XL H1.
The 24f simulation is very, very good. It is almost indistinguishable from true 24p but there are indeed slight interlace artifacts. HERE IS PROOF, from the "XLH1carspassingstatic.m2t" clip:
Note: These were saved as uncompressed JPEGs (quality 12 out of Photoshop).
Note the aliasing. It is slight, but present. I believe Canon did as good a job as anyone could with interlace CCDs, but it's still not perfect.
And to answer the other guy's question about capturing the same image in both fields, there's no way. An interlace CCD will run at 48 Hz (or 60), capturing images into fields at 1/48 sec increments. Nothing you can do about that. It's the design of the CCD.
My hypothesis is that Canon does some type of intelligent blending between fields, similar to deinterlacing tools in post. The difference is that the in-camera deinterlacer (or blender) is operating on raw data before HDV compression.
I hope this clears up some of those issues.
Posted 26 January 2006 - 04:54 PM
But that's the thing Thomas.....No other gear "WILL" read 24f it except another XL-H1. *smile* Which gives indication it is in fact 24p recorded. Additionally, programs like Vegas does not add pulldown when reading the files and that confirms it even more. it's crazy but it's true. You may not believe me, but trust me, that's the case. otherwise, the Sony would read the 24f footage just fine....but it does not.
There may be a mode to write it without pulldown, but I have not seen any footage that way, only 60i with 3:2 pulldown. I doubt it, however, as this would not conform to the HDV spec and would not be readable by any other post gear except another XL H1.
However, I tend to beleive you are correct, and so is David Mullen. I was wrong before about recording to tape. The 24f "recorded to tape" is in fact 24p with 3:2 pulldown as David said it was. At least I think it is. I think it is 60i in all forms on tape, just like the XL2, DVX100, SDX900 or any other 24p camera that needs to conform to a 29.97 tape spec. It is however 23.976 progressive coming directly out of the firewire....no pulldown added. That's where I got mixed up when talking with David in the other post.
Also, why would you use Kaku's old footage as an example? It was from a pre-production model XL-H1 he got almost 6 months ago! dvinfo has removed that as a sticky and doesn't even use that as example footage anymore. Moreover, some of Kaku's footage was at question to what format it was in and set on. 24p footage was 29.97 when we all now the camera spits out 23.976 in 24f mode. So that footage is not quantified as a good example to use. Thomas, why not use more modern PRODUCTION MODEL footage to illustrate 24f? From cameras that people have purchased and own. There is plenty of it on the net.
Actually...there is footage being shot at SUNDANCE 2006 THIS WEEK (right now) with the Canon XL-H1 that you can use for illustration purposes...this is from an over-the-counter camera final production camera that we can all go get at your local pro video store. From the reports, everybody is BLOWN AWAY from the 24p HD stuff it is producing. The footage aired on Sundance channel this past Monday and WILL BE AIRING MORE 24F FOOTAGE TONIGHT AS WELL. However I think tonites showcase was captured in SD and not HD s monday's stuff was. Find more about that here:
I think that's a bit more fair, wouldn't you agree?
As far as Aliasing goes...heck...even the HVX200 with it's progressive scanning chips has aliasing. It comes from the artificial edge enhancement these cameras uses when the DETAIL setting is turned up too much. Kaku did not dial the camera in like it should have. He just pulled it out the box, pointed and shot. We have already done a camera shootout and many many many other people have already checked this out, and the question of 24f recording 24p is no longer a question. It is. PRODUCTION MODEL 24f footage has been examined, disected, scrutinized, inspected, evaluated, microscoped and given a full autopsy by many people already. The XL-H1 records every bit of 24p as the XL2 does (actually, it looks better). In High Definition, 24f resolves more resolution then any other camera available for under $60,000. And it looks better then any other camera too. Ofcourse that's a subjective opinion, but isn't everything in Video? *smile*
I agree with you, I do not beleive the CCD's are scanning both fields simultaneously. But then again nobody knows the 'design of the CCD' either. It's all speculative. Some people think its a native 1440x1080 pixel shifting chip. Others say its a full 1920x1080 pixel shifting chip. Crazy I know...but the point is, we just don't know for sure. These companies got spin on everything they do. It's ridiculous. *smile*
I guess what I'm saying is some of the issues you are attempting to clear up are already clear....just the results are different I suppose.
Posted 26 January 2006 - 05:34 PM
I would be more than happy to look at newer footage, if you can send me a link to some clips directly out of the camera. I would like to experiment simply for my own knowledge as well.
Thomas, why not use more modern PRODUCTION MODEL footage to illustrate 24f? From cameras that people have purchased and own. There is plenty of it on the net.
Or better yet, just give me your camera.