|
|||||||||
|
Thread Tools | Search this Thread |
July 4th, 2005, 01:08 PM | #16 | |
RED Problem Solver
Join Date: Sep 2003
Location: Ottawa, Canada
Posts: 1,365
|
Quote:
I will be putting in options for 4:2:0 into my filter packages. I don't have any clips, but that's what the free demo is for so you can try on your own footage. Graeme
__________________
www.nattress.com - filters for FCP |
|
July 4th, 2005, 04:25 PM | #17 |
Regular Crew
Join Date: Sep 2001
Location: California
Posts: 667
|
[QUOTE=Graeme Nattress] I will be putting in options for 4:2:0 into my filter packages.
Will there be an option to make a 4:2:2 version on output? I can't wait to try your plugins, I have heard nothing but praise about them! Thanks Graeme! |
July 4th, 2005, 04:48 PM | #18 |
RED Problem Solver
Join Date: Sep 2003
Location: Ottawa, Canada
Posts: 1,365
|
Yes, the 4:2:0 option will try to boost the chroma to 4:2:2.
Graeme
__________________
www.nattress.com - filters for FCP |
July 4th, 2005, 06:30 PM | #19 | |||
Major Player
Join Date: Nov 2004
Location: LI, NY
Posts: 274
|
Quote:
Quote:
http://www.koepi.org/XviD-1.0.3-20122004.exe Quote:
http://www.megaupload.com/?d=20P9EICE While it gets rid of the pulldown perfectly, fast motion seems jerky. The reason behind that is that there is no motion blur. Watch the clip and look for the man walking in the background, he's perfectly smooth. Then watch the motorbikes, jerky. Do a frame by frame on the bikes and you can see how this mode would be great for slow-mo on fast moving objects...maybe. I'd have to do some tests with that. The clip I tested on consisted of myself walking around a room touching certain objects and even jogging. That came out fine since I was not moving 20+ mph. |
|||
July 5th, 2005, 10:01 AM | #20 | |
Regular Crew
Join Date: Sep 2001
Location: California
Posts: 667
|
Excellent!
Thanks Graeme pappas Quote:
|
|
July 7th, 2005, 05:41 PM | #21 |
New Boot
Join Date: Mar 2005
Location: Los Angeles, CA
Posts: 11
|
Thoughts on this article?
Graeme/Douglas/all,
This was the original article that I remembered reading regarding Magic Bullet and the "film look". http://thecarpark.net/products_magicbullet2_HDV.htm# Although he doesn't claim it converts from 60i to 24p, Christopher Kenworthy seems to believe that Magic Bullet is excellent at giving video that "film look" through blurring, etc. Since, I don't understand it, Graeme, can you please explain the difference between your product and Magic Bullet if both give the "film look"? Yours gives it apparently through 60i to 24p and Magic Bullet through some other means. Thanks in advance, Douglas |
July 7th, 2005, 06:56 PM | #22 |
RED Problem Solver
Join Date: Sep 2003
Location: Ottawa, Canada
Posts: 1,365
|
Really, the only thing magic about the bullet is their marketing budget!
Yes, I have my own proprietary algorithms that I believe produce an excellent look. But really, the major difference is that my plugins are significantly cheaper and you get tech support direct from me, the person who writes the code. MB for Editors does not do the 24p thing, or de-interlace. The AE version does, but that's $1000, not $300. Oh, and free upgrades for Film Effect customers from day one. Hopefully free upgrades won't end, but people who've bought from V1.0, will be getting a free upgrade to V2.5 which will be released this month, and I can't really get any fairer than that. Graeme
__________________
www.nattress.com - filters for FCP |
July 8th, 2005, 03:12 PM | #23 |
Trustee
Join Date: Oct 2002
Location: Arlington VA
Posts: 1,034
|
I did this using VirtualDub and obtained near perfect results.
The procedure was to convert the M2T to AVIs using MidVid MJPEG. I'd like to do it with DVCProHD but can't find a good cheap codec (any suggestions?) In virtualdub, I set it to inverse telecine with a 0 field offset to 23.976 fps. Then I added a 1-pixel blur filter, and finally I downed it to 1280x720 (since the blur does cause some resolution loss). I think the blur causes a loss to slightly less than true 720p quality but certainly much better than 480. The footage is progressive and gorgeous and is virtually indistinguishable from 24p from a Panasonic AG-DVX100A when downed to 480p, excepd that the picture quality is immensly superior. I will try to post a clip this afternoon. |
July 14th, 2005, 06:56 PM | #24 |
Major Player
Join Date: Mar 2005
Location: Ventura, California, USA
Posts: 751
|
What is the purpose of the 1-pixel blur?
|
July 14th, 2005, 08:05 PM | #25 |
Major Player
Join Date: Nov 2004
Location: LI, NY
Posts: 274
|
Wouldn't matter anyway, you'd get blended fields. Might as well try to pass those off as motion blur.
|
July 14th, 2005, 09:03 PM | #26 |
Trustee
Join Date: Mar 2004
Location: Milwaukee, WI
Posts: 1,719
|
converting 60i to 60p will give much better results when going to 24p.
with 30i or 30p the 24p frames happen within every 1.25 of the 30p frames. So only every other 4th frame will land on a real frame. With 60p the 24p frames happen within every 2.5 of the 60p frames. That means only every other 24p frame is interpolated. The in between frames are real frames from the 60p. There is no more quality loss from going to 60p either since you have to deinterlace anyways. When you deinterlace 60i to 30p you throw away one whole field of data or basically half the frames. If you seperate the 60i into 60 half height frames and then scale up to full height you get 60p that will look just as good as 30p. The best would be to convert to 120p. 120p gives a 24p frame exactly every 5 frames. Just take every 5th frame from your 120 sequence and you have a perfect 24p. Of course going to 120p can take a long time. I am working on a program to do a high quality 60p and convert that to 24p. Every other frame will be a perfect deinterlaced frame. The in between frames will be blurry or interpolated. I am still working on how to best create the in between frame. If I can make a time shifter that is fast enough I may even go the 120p route. |
July 14th, 2005, 10:32 PM | #27 |
Trustee
Join Date: Oct 2002
Location: Arlington VA
Posts: 1,034
|
The way it works is:
A1 A2 B1 B2 C1 C2 D1 D2 E1 E2 becomes A1+A2+blur = A_P B1+B2+blur = B_P C2+D1+blur = C_P D2+E1+blur = D_P You have to blur because the two interlaced fields happen at different points in time. Also known as "Blending fields" which VirtualDub doesn't do unless you tell it to. But then to compensate for the blur, you go down to 720p so it all works out. And then you get beautiful 24p footage which only the most discerning eyes would know was artificial. |
July 15th, 2005, 12:14 AM | #28 | |
Major Player
Join Date: Nov 2004
Location: LI, NY
Posts: 274
|
Hmm, Thomas I'll try your method and see how that works.
Quote:
http://www.uploadhouse.com/images/625620499020.JPG You're not creating a true progressive source to work with. You're basically blending. The motion may appear ok to your eyes, but you are creating a mess with the frames. EDIT: 60i to 24p - 119.88 to 23.976 http://www.megaupload.com/?d=283SRN7Y 60i to 30p - Deinterlaced http://www.megaupload.com/?d=2890NCTV 60i to 60p - Resized Fields http://www.megaupload.com/?d=27BZ47GT I've uploaded all three so we can compare the motion in each. The sample footage is only 10sec. I'll do a longer test tomorrow. Last edited by Kyle Edwards; July 15th, 2005 at 12:48 AM. |
|
July 15th, 2005, 01:38 AM | #29 |
Trustee
Join Date: Mar 2004
Location: Milwaukee, WI
Posts: 1,719
|
What program are you using to convert to 120p?
|
July 15th, 2005, 07:14 AM | #30 |
Major Player
Join Date: Mar 2005
Location: Prague, Czech Republic
Posts: 500
|
Graeme & Co,
Once convert 1080i60 to 24p, using your, other top software, what we getting subjectively as result, is overall quality close to 720p, 1080p, somewhere in between, lower than 720p? What I'm driving at, if lenses are same, would picture quality be better on 720p24 HDV, 720p24 DVCPROHD with 960x720 recorded pixels, or Sony HDV converted from 50-60i to 24-25p? David Newman of CineForm already said that 720p24 HDV codec means theoretically better quality than DVCPROHD at 24p, which is 40 Mbps. How close is Sony HDV codec to HDCAM codec at 1080i? Compressin on non-moving complex image is about same (significantly less than on DVCPROHD), once movement starts, compression increases but eye is less sensitive to movement so it may perceive it as natural blur. Radek |
| ||||||
|
|