What's Your FX1 "24fps" Solution? at DVinfo.net

Peter Moretti · July 23rd, 2007, 07:41 PM

I've read that CineFrame reduces HD resolution quite a bit, so how are you acheiving a 24p look with your FX1? And please include what shutter setting you're using.

Thanks much!

Brandon Freeman · July 24th, 2007, 07:52 AM

Hm. I haven't found that CineForm reduces resolution. My workflow, when converting from 60i to 24p is the following:

Shoot at 1/60 shutter speed at all times.

Capture .m2t in Vegas 7 or Connect HD.

Send the .m2t to DVFilm Maker (www.dvfilm.com) and use the Connect HD codec.

From there, I edit.

Ben Winter · July 24th, 2007, 10:05 AM

I am the same, although I capture to Cineform .avi before 24p conversion and then export out to it as well. I suppose .m2t might have less generation loss, I'll have to try that sometime.

If I'm doing a quick project that I don't want to invest too much time in, I just shoot Cineframe24 and use Cineform's pulldown converter to get me a real 24p file to work with. Doesn't look as good (WAY too juddery) but it's good for on-the-fly or 24/48 hr film contests.

Ian Holb · July 24th, 2007, 10:21 AM

Ok, my advice won't be as helpful but the easiest I've found is to upgrade to the Sony Z1U from the FX1 for about $900 more. Shoot in Cineframe 25 with a shutter speed of 50, or shoot in 50i and de-interlace in post. Depending on the quality of the de-interlacing software (adaptive motion or not), it can look better than Cineframe25 although the workflow is a tad more convoluted. Final step is to conform to 23.976fps.

There really is no 'easy' way to get 24fps from the Sony FX1 NTSC version without expensive equipment or great deal of time.

The best way, coming from Apple's solution, is to edit Cineframe30 or 60i material as usual, and then render out the final project using Compressor 3's frame rate control. Compressor uses Shake optical flow technology to intelligently reconstruct a true 24P source from your 60i material. The downside is crazy rendering times in the neighborhood of hours and days.

Another way is to re-rate your footage using Adobe After Effects v7 or CS3. It also has an optical flow retiming method that is very high quality.

Brandon Freeman · July 24th, 2007, 11:35 AM

Well, yes, if you've got the ability to upgrade to the Z1U, that is an option (that's actually what I do with my Z1U -- shoot in PAL in CineFrame25 and work in a 1280x720 timeline since that's about what the resolution is at after the deinterlacing). Even though I do get reduced resolution with CineFrame25 (it's not terrible at all for my applications), I don't get nearly as many artifacts as when I shoot interlaced. The lines cause a lot more artifacting, and the softer, psuedo-progressive image seems to encode easier, so that's another thing to consider.

But I WOULD NOT say to upgrade to a Z1U now. That's more or less taking a side step if you want to actually get something newer, that can do progressive 24p. You could go to the V1U, the Canon A1, or the Panasonic HVX200. However, I would say hold onto your pennies and wait for the next generation of HD/V -- I think that the new XDCAM EX series (not to mention higher end revolutions like the RED) is going to bring a whole new meaning to "affordable digital cinematography".

Serena Steuart · July 24th, 2007, 07:26 PM

The best way to get to 24/25P is to shoot native (60i/50i) and to convert in post. Cineframe 25 does produce a good result, but processing 50i with DVFilmMaker is superior in all except time involved. Cineform does a good job of conversion, somewhat less than DVFilmMaker but satisfactory to many and one pass less.

If you want to edit m2t files then Cineframe is the only way to go unless you are going to hand the final conversion to a post facility.

Otherwise for best results the workflow is:
shoot camera native format
convert to avi using Cineform NEOHDV or NEO-HD (Mac) or Prospect HD (PP)
process avi through DVFilmMaker
edit and render at 24P

There are various tunes to be played on the workflow and how you do it (or which software) depends on the results you want and the time available for processing.

Brandon Freeman · July 24th, 2007, 08:09 PM

The only issue, Serena, with de-interlacing in post, is the added artifacts that become horribly exposed. Interlacing strains the MPEG codec. I've found that the deinterlacing in the camera produces a much cleaner image.

Serena Steuart · July 25th, 2007, 07:27 AM

" deinterlacing in the camera produces a much cleaner image"

Well the camera doesn't de-interlace, it just chucks one field and interpolates. So that reduces vertical resolution, but if that satisfies then that is the simplest solution. The issue about interlace artifacts is influenced very strongly by the display device. If the display is interlaced then really there are no artifacts. The matter becomes significant when interlaced is shown on a progressive display (such as an LCD screen). These disturbed me greatly because I'd used film until very recently and reckoned that all these video people saying "no problem" had been brainwashed by having never seen a decent image. But it was all in the display device and if that will display interlaced as interlaced, there are few artifacts.
If you are sourcing your data from a computer hard disk (such as DVI input to a projector) then interlace artifacts are truly horrible. This is easily fixed while retaining resolution by intelligent de-interlacing using DVFilmMaker, which is very good for relatively cheap software. More expensive software does a better job. If you haven't looked at the technology, then visit that website.

Brandon Freeman · July 25th, 2007, 08:33 AM

I'm not referencing the interlace artifacts (the hair comb effect) but the compression blockiness that the interlacing artifacts cause on the MPEG LONG GOP. Once de-interlaced in DVFilm Maker (which I do own), these blocks become all the more visible, whether on a computer or on a TV. This is why HDV (specifically the 1080i kind) has been called "messy" and "blocky" -- I just don't think it can handle all the motion from interlacing lines.

Here's a 60i HDV still that was de-interlaced using DVFilm Maker. Note the very dirty, blockiness where there is motion. This is not DVFilm's fault -- this is the MPEG codec unable to handle the high motion with interlacing artifacts.

http://www.hall-e-woode.com/hvrz1ute...ced_DVFilm.jpg

And even if one can't see the blockiness when the picture is in motion, they make the picture feel digitized. I'd rather have a soft image resembling a lower quality film stock than a higher resolution that looks blocky and screams video.

So, while resolution is dropped in CineFrame25, so are the compression blocks. And while CF25 is not acceptable for 1080i/p display, it works well (I've found) on a 720p timeline, and obviously even better on an SD timeline.

Ben Winter · July 25th, 2007, 10:26 AM

Quote:

Originally Posted by Brandon Freeman

So, while resolution is dropped in CineFrame25, so are the compression blocks. And while CF25 is not acceptable for 1080i/p display, it works well (I've found) on a 720p timeline, and obviously even better on an SD timeline.

The reason you see less blocking with Cineframe is because it literally drops frames completely. You won't eliminate the blocky artifacting with Cineframe, because it's still MPEG compression, you'll just lessen it because you're literally seeing less information. The only difference is that Cineframe25 just samples frames at 1/25th intervals. DVFilm uses all interlaced frames of 60i to create 1/48 motion blur, so yes, the motion blocks will still be there.

Shooting 24p means treating your camera like a film camera, which means slow, deliberate movement. Jerking around your camera too suddenly can be very difficult to watch due to 24p motion judder, especially on larger screens. This is the only time I ever see MPEG compression blocks manifest themselves, so I really never run into this issue. If you shoot handheld, jerky video on a regular basis you're better off shooting DV or less compressed video ala HVX.

Ervin Farkas · July 25th, 2007, 01:06 PM

Brandon, what was the lighting on that shot? Fluorescent? Do you remember your camera settings by any chance? Was the camera on tripod?

This particular image seems to have lots of "issues"...

Brandon Freeman · July 25th, 2007, 03:58 PM

Yes, there are multiple issues, as I was shooting to expose another issue -- the stairstepping issue on the light that drove me crazy. I shot hand held at 60i, 1/60th shutter, on one knee. I determined that, no matter what, I could not get rid of nearly horizontal objects stairstepping, no matter which way I de-interlaced -- and determined that the camera's sharpening needed to be turned to 0. But as I was doing the tests, I found the blocks on lots of motion were insane. Definitely turned me off to interlaced HDV.

Ben, exactly. That's why I use Cineframe in PAL -- to lessen the impact of encoding. Screw the film look, I do it because I get less encoding on each moving image. The same reason that I will use 24f modes almost exclusively when the church converts to a nearly exclusive Canon XH-A1 system sometime next year (crossing my fingers). As far as filming with the rules of film (subtle movements), generally, that's how I shoot. But sometimes, when you're shooting a church event, you gotta say, "To hell with it." (Funny, you know, being in a church service and all. :) )

Ian Holb · July 25th, 2007, 05:45 PM

Quote:

Originally Posted by Brandon Freeman

But I WOULD NOT say to upgrade to a Z1U now.

After I wrote that, I thought instead of the Z1U get a Canon XH-A1. That might be the easier and best solution although it wouldn't help him if he has stacks of 60i footage he's already shot.

Still, my best advice is taking 60i footage into Compressor 3 and have that program spit out true 24P (complete with subpixel accurate re-timing and ultra high quality motion-compensated de-interlacing)

Peter Moretti · July 25th, 2007, 07:53 PM

Quote:

Originally Posted by Ian Holb

...

Still, my best advice is taking 60i footage into Compressor 3 and have that program spit out true 24P (complete with subpixel accurate re-timing and ultra high quality motion-compensated de-interlacing)

Ian,

I live in a pc production world.

Are you aware of any pc-based deinterlacers comparable to Apple's Compressor 3? FWIW, the NLE's I have are Sony Vegas and Avid Express Pro.

Thanks a lot!

Brandon Freeman · July 26th, 2007, 07:47 AM

Quote:

Originally Posted by Peter Moretti

I've read that CineFrame reduces HD resolution quite a bit, so how are you acheiving a 24p look with your FX1? And please include what shutter setting you're using.

Thanks much!

Oh...my gosh. I'm so sorry. I should have read this first post clearer. I thought you were saying CineForm (the codec) reduces resolution.

My bad.

Yes, CineFrame does reduce resolution, but... (fill in all that discussion that took place already) ...just make sure you don't use CineFrame24, 'cuz it's crrrrrap!

:)

July 23rd, 2007, 07:41 PM	#1
Peter Moretti Trustee Join Date: Mar 2007 Location: Sherman Oaks, CA Posts: 1,259	What's Your FX1 "24fps" Solution? I've read that CineFrame reduces HD resolution quite a bit, so how are you acheiving a 24p look with your FX1? And please include what shutter setting you're using. Thanks much!

July 24th, 2007, 07:52 AM	#2
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	Hm. I haven't found that CineForm reduces resolution. My workflow, when converting from 60i to 24p is the following: Shoot at 1/60 shutter speed at all times. Capture .m2t in Vegas 7 or Connect HD. Send the .m2t to DVFilm Maker (www.dvfilm.com) and use the Connect HD codec. From there, I edit. Last edited by Brandon Freeman; July 24th, 2007 at 07:52 AM. Reason: forgot to include DVFilm's website

July 24th, 2007, 10:05 AM	#3
Ben Winter Trustee Join Date: Jun 2005 Location: Hollywood, CA Posts: 1,675 Images: 1	I am the same, although I capture to Cineform .avi before 24p conversion and then export out to it as well. I suppose .m2t might have less generation loss, I'll have to try that sometime. If I'm doing a quick project that I don't want to invest too much time in, I just shoot Cineframe24 and use Cineform's pulldown converter to get me a real 24p file to work with. Doesn't look as good (WAY too juddery) but it's good for on-the-fly or 24/48 hr film contests. __________________ BenWinter.com

July 25th, 2007, 01:06 PM	#11
Ervin Farkas Inner Circle Join Date: Aug 2005 Location: Atlanta/USA Posts: 2,515	Brandon, what was the lighting on that shot? Fluorescent? Do you remember your camera settings by any chance? Was the camera on tripod? This particular image seems to have lots of "issues"... __________________ Ervin Farkas www.AtlantaLegalVideo.com

July 24th, 2007, 10:21 AM	#4
Ian Holb Regular Crew Join Date: Nov 2006 Location: Canada Posts: 127	Ok, my advice won't be as helpful but the easiest I've found is to upgrade to the Sony Z1U from the FX1 for about $900 more. Shoot in Cineframe 25 with a shutter speed of 50, or shoot in 50i and de-interlace in post. Depending on the quality of the de-interlacing software (adaptive motion or not), it can look better than Cineframe25 although the workflow is a tad more convoluted. Final step is to conform to 23.976fps. There really is no 'easy' way to get 24fps from the Sony FX1 NTSC version without expensive equipment or great deal of time. The best way, coming from Apple's solution, is to edit Cineframe30 or 60i material as usual, and then render out the final project using Compressor 3's frame rate control. Compressor uses Shake optical flow technology to intelligently reconstruct a true 24P source from your 60i material. The downside is crazy rendering times in the neighborhood of hours and days. Another way is to re-rate your footage using Adobe After Effects v7 or CS3. It also has an optical flow retiming method that is very high quality.

July 24th, 2007, 11:35 AM	#5
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	Well, yes, if you've got the ability to upgrade to the Z1U, that is an option (that's actually what I do with my Z1U -- shoot in PAL in CineFrame25 and work in a 1280x720 timeline since that's about what the resolution is at after the deinterlacing). Even though I do get reduced resolution with CineFrame25 (it's not terrible at all for my applications), I don't get nearly as many artifacts as when I shoot interlaced. The lines cause a lot more artifacting, and the softer, psuedo-progressive image seems to encode easier, so that's another thing to consider. But I WOULD NOT say to upgrade to a Z1U now. That's more or less taking a side step if you want to actually get something newer, that can do progressive 24p. You could go to the V1U, the Canon A1, or the Panasonic HVX200. However, I would say hold onto your pennies and wait for the next generation of HD/V -- I think that the new XDCAM EX series (not to mention higher end revolutions like the RED) is going to bring a whole new meaning to "affordable digital cinematography".

July 24th, 2007, 07:26 PM	#6
Serena Steuart Major Player Join Date: Jun 2006 Location: Studio Alnitak, Melbourne, Victoria, Australia Posts: 640 Images: 3	The best way to get to 24/25P is to shoot native (60i/50i) and to convert in post. Cineframe 25 does produce a good result, but processing 50i with DVFilmMaker is superior in all except time involved. Cineform does a good job of conversion, somewhat less than DVFilmMaker but satisfactory to many and one pass less. If you want to edit m2t files then Cineframe is the only way to go unless you are going to hand the final conversion to a post facility. Otherwise for best results the workflow is: shoot camera native format convert to avi using Cineform NEOHDV or NEO-HD (Mac) or Prospect HD (PP) process avi through DVFilmMaker edit and render at 24P There are various tunes to be played on the workflow and how you do it (or which software) depends on the results you want and the time available for processing.

July 24th, 2007, 08:09 PM	#7
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	The only issue, Serena, with de-interlacing in post, is the added artifacts that become horribly exposed. Interlacing strains the MPEG codec. I've found that the deinterlacing in the camera produces a much cleaner image.

July 25th, 2007, 07:27 AM	#8
Serena Steuart Major Player Join Date: Jun 2006 Location: Studio Alnitak, Melbourne, Victoria, Australia Posts: 640 Images: 3	" deinterlacing in the camera produces a much cleaner image" Well the camera doesn't de-interlace, it just chucks one field and interpolates. So that reduces vertical resolution, but if that satisfies then that is the simplest solution. The issue about interlace artifacts is influenced very strongly by the display device. If the display is interlaced then really there are no artifacts. The matter becomes significant when interlaced is shown on a progressive display (such as an LCD screen). These disturbed me greatly because I'd used film until very recently and reckoned that all these video people saying "no problem" had been brainwashed by having never seen a decent image. But it was all in the display device and if that will display interlaced as interlaced, there are few artifacts. If you are sourcing your data from a computer hard disk (such as DVI input to a projector) then interlace artifacts are truly horrible. This is easily fixed while retaining resolution by intelligent de-interlacing using DVFilmMaker, which is very good for relatively cheap software. More expensive software does a better job. If you haven't looked at the technology, then visit that website.

July 25th, 2007, 08:33 AM	#9
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	I'm not referencing the interlace artifacts (the hair comb effect) but the compression blockiness that the interlacing artifacts cause on the MPEG LONG GOP. Once de-interlaced in DVFilm Maker (which I do own), these blocks become all the more visible, whether on a computer or on a TV. This is why HDV (specifically the 1080i kind) has been called "messy" and "blocky" -- I just don't think it can handle all the motion from interlacing lines. Here's a 60i HDV still that was de-interlaced using DVFilm Maker. Note the very dirty, blockiness where there is motion. This is not DVFilm's fault -- this is the MPEG codec unable to handle the high motion with interlacing artifacts. http://www.hall-e-woode.com/hvrz1ute...ced_DVFilm.jpg And even if one can't see the blockiness when the picture is in motion, they make the picture feel digitized. I'd rather have a soft image resembling a lower quality film stock than a higher resolution that looks blocky and screams video. So, while resolution is dropped in CineFrame25, so are the compression blocks. And while CF25 is not acceptable for 1080i/p display, it works well (I've found) on a 720p timeline, and obviously even better on an SD timeline.

July 25th, 2007, 03:58 PM	#12
Brandon Freeman Major Player Join Date: Feb 2007 Location: Boise, ID Posts: 226	Yes, there are multiple issues, as I was shooting to expose another issue -- the stairstepping issue on the light that drove me crazy. I shot hand held at 60i, 1/60th shutter, on one knee. I determined that, no matter what, I could not get rid of nearly horizontal objects stairstepping, no matter which way I de-interlaced -- and determined that the camera's sharpening needed to be turned to 0. But as I was doing the tests, I found the blocks on lots of motion were insane. Definitely turned me off to interlaced HDV. Ben, exactly. That's why I use Cineframe in PAL -- to lessen the impact of encoding. Screw the film look, I do it because I get less encoding on each moving image. The same reason that I will use 24f modes almost exclusively when the church converts to a nearly exclusive Canon XH-A1 system sometime next year (crossing my fingers). As far as filming with the rules of film (subtle movements), generally, that's how I shoot. But sometimes, when you're shooting a church event, you gotta say, "To hell with it." (Funny, you know, being in a church service and all. :) )