Cineframe 30/Cineframe 25 resolution loss vs. 60i
by Barry Green
on
Dec 20, 2004 at 7:46:05 pm
I know there has been lots of speculation about CineFrame 30 and how it may employ a "smart" de-interlacer that would only cause resolution loss in areas of motion, and no resolution loss in static areas.
Unfortunately, it isn't so. I've extracted the resolution sections from a CamAlign chart, shot both in 60i and CF30. The CF30 shows substantial resolution loss, and this is coming off a completely locked-down still shot. Looks like there's no smart de-interlacing going on.
Interestingly, Steve Mullen had hypothesized that due to row-pair summation and low-pass filters, the FX1 should deliver an "effective" 820 lines of resolution, and that seems to have been a pretty darn good guess. In this chart I'd say that it cleanly resolves at 700, and things get a little blurred together at about 750, so for a guess of 820 that's *very* close.
In interlaced mode the 1080i of the HDV delivers about 750 lines of resolution. In CF30 mode it drops to about maybe 575. Which is close enough to say that CF30 is pretty much a straight frame-drop de-interlace.
For the pictures, one is a 300% blow-up in Photoshop that demonstrates the convergence of lines on the resolution portion of the chart.
[img]http://www.icexpo.com/FX1-60i-vs-CF30-res.jpg[/img]
For those who want to examine the raw extraction from the camera, here's a bitmap extraction of the uncompressed frames from the .m2t (no CineForm intermediary codec):
[img]http://www.icexpo.com/FX1-60i-vs-CF30.BMP[/img]
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 20, 2004 at 7:50:46 pm
I guess I can't edit posts. Let me add for clarification: when I say "straight field-drop", that doesn't mean that resolution is cut in half from the interlaced version (as obviously, it'd have to be down around 380 for that to be true). Instead, I'm saying that it looks like in CF30 the camera skips the field-blending and low-pass filter and lets the full raw resolution of one field pass through. That would give us a theoretical 540 lines, which is within the margin of error of the 575 that the chart is showing.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 21, 2004 at 3:49:33 am
It depends on what you mean by "compare". In some ways they can't compare at all -- the FX1 is high-definition, and when viewed on a high-def monitor it's in another league entirely.
In other ways the DVX and XL2 are quite superior to the FX1 -- 24P motion, light sensitivity, audio, etc.
Different tools for different uses. If you want to make high-def, the FX1 is the only practical choice right now. But if you want to make DVD's for standard-def release, the DVX and XL2 are better overall cameras for that purpose. Like a spoon is good for soup but doesn't do much for steak, and a fork is better for steak but doesn't do so good for soup.
I've posted lots of photos and comparisons of DVX vs. XL2 vs. FX1, when used for the purpose of making DVD's, on some of the other forums.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Moses on Dec 21, 2004 at 4:15:50 am
I guess I meant in pure resolution chart terms.
I'm kinda a novice when it comes to resolution, so I just assumed that DV cameras turned out 480 lines of resolution, which isn't that far away from your assesment of the FX1 in CF mode. But after looking at your test I'm wondering if the XL2 and DVX actually deliver 480 lines? I've read conflicting things online, and your test looks like something I can trust.
PS. Am I right in assuming that a smart de-interlacer in post could turn 60i of a STILL resolution chart into 30P without sacrificing resolution?
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 21, 2004 at 4:32:29 am
Ah. Well, in pure resolution chart terms, the FX1 wins easily. It's high-def. It puts down about 775 lines of vertical resolution, versus a maximum of 480 to 540 (depending on whether you're talking "TV Lines" or actual scan lines) for the DVX and XL2.
In horizontal resolution the FX1 delivers somewhere between 800 and 900 lines, or about double what the DVX and XL2 can do.
Even in CF30 mode the FX1 is putting out about 550 lines, competitive with the DVX and XL2, but still having twice the horizontal res.
The thing is, the DVX and XL2 can shoot progressive-scan, and in progressive-scan mode they can deliver the full resolution of the frame without the row-pair summation and vertical blur that is employed in interlaced mode. In interlaced mode the DVX and XL2 do about 360 lines, and the FX1 does about 775 even though it has 1080 pixels to work with).
In any contest of resolution, when displayed on a high-def monitor, the FX1 will handily win. When displayed on a standard-def monitor is another story, and for DVD release they deliver about the same, resolution wise; on the downconversions I've been doing and comparing between the cameras the DVX and XL2 look sharper on DVD than the FX1 does.
As a DV camera, the FX1 does not excel -- it can come close to the others, but doesn't have nearly the image control, light sensitivity, frame rate or audio of the DVX and XL2. As a high-def camera, the FX1 is (currently) unparalleled, at this price point. If you like the look of interlaced HD, and can display it on an HDTV, the FX1 is clearly superior to any DV camera.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 21, 2004 at 4:38:57 am
[Moses]"PS. Am I right in assuming that a smart de-interlacer in post could turn 60i of a STILL resolution chart into 30P without sacrificing resolution?"
Forgot to answer this part. Yes, a properly intelligent smart de-interlacer should be able to turn a still shot from 60i to 30p with no resolution loss.
However, the very process of shooting the chart at 60i is causing resolution loss. Between the row-pair summation, the low-pass filtering, and the kell factor, you'll see a much lower resolution picture coming from an interlaced camera, as opposed to a progressive-scan camera. The XL2 and FX2 in 24P mode can turn out a picture with 480 scan lines of resolution, but when you put them into interlaced mode the best they can do is about 360 lines (NTSC numbers here). Interlaced cameras blur the fields together to minimize interline flicker.
So if you're shooting your res chart in 60i mode, you'll get the highest res that the interlaced camera can deliver, which in the case of the FX1 is about 775 lines (even though the pixel count goes up to 820). Then when wanting to make a 30P simulation of that, you could use a smart de-interlacer (such as DVFilm's Maker software), which could theoretically preserve all that resolution (and, in which case, would look no different than the 60i shot).
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by karl on Dec 21, 2004 at 10:01:41 am
yeh thanks barry ive been asking this question all along at the cow. It's nice to see someone do a resolution test in cineframe mode.
I was considering the FX for a low budget film, but if the cinemode is really the same res as an XL2 then the difference isnt as hugely clear cut as I thought. Sure it is better vertical resolution and I could go the software de-interlaced approach to ger more horizontal res. However, something tells me this is scratching the surface of HDV and there will be big improvements later this year.
Im becoming more inclined to wait and see how it develops by the summer, and what JVC have to offer.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 21, 2004 at 6:14:14 pm
[karl]"but if the cinemode is really the same res as an XL2 then the difference isnt as hugely clear cut as I thought"
Well, let's be clear again: in vertical pixel count a PAL XL2 in 25P/THIN is going to deliver the same # of discrete scan lines as an FX1 in CF25, but the FX1 is still going to have twice as many (and interpolated to 2.5 times as many) horizontal pixels. So I wouldn't say they're the "same res", the FX1 is still going to look much higher resolution (when viewed on an HDTV).
When viewed on DVD, no there won't really be any resolution advantage to the FX1. But when on an HDTV the FX1 will still enjoy a significant resolution advantage. Just want to make sure what I'm saying isn't misunderstood...
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by karl on Dec 21, 2004 at 7:02:17 pm
no i realise that, but at large projections I think the horizontal res is very important too.
I really dont think it will be too long until someone brings out a true progressive scan HDV camera and maybe it's worth waiting until some announcments in the new year.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 21, 2004 at 6:13:00 pm
"Instead, I'm saying that it looks like in CF30 the camera skips the field-blending and low-pass filter and lets the full raw resolution of one field pass through."
Are you thinking the CCD does not row-sum in the CF modes?
It's true that that is a way of gaining maximum field V rez. But, then sensitivity should be about 6dB less in CF modes. Is it?
I think in the CF modes, interlace video is always input frpm the CCDs. Then, for static OBJECTS the full V rez should be obtained. For OBJECTS in motion, the V rez -- which you can never measure -- is about 410-lines.
IF you are measuring V rez (static) at about 600-lines -- then it seems far more likely Sony has limited static V rez so the difference between static and dynamic is less obvious. (See the other posts on motion issues.)
In short, you really need to consider that the deinterlacing is part of MPEG-2 encoding. Thus, it works nothing like the other cameras.
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 21, 2004 at 6:23:16 pm
Sorry, you HAVE measured V rez in CF30. I'm reading mail, while in India backwards. :)
Can you check light sensitivity between CF30 and 60i.
I've always though Sony was very clever to always use 60i because it did keep sensitivity constant.
There is another possibility for the 575. Remember how with the DVX100 you can turn ON 2 levels (in the 100A) of blending in the 24A mode. As I remember, if you were going to film you did NOT turn on the filter. However, if you were going to display on interlace devices -- you needed to filter to less Vrez to prevent line-twitter and Morie patterns. (
Or. maybe if you were going to an interlace format like to a DVD.)
So maybe thats the cause for the loss.
Does the Z1 have a Full V rez menu item?
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 21, 2004 at 6:27:52 pm
[Steve Mullen]"Can you check light sensitivity between CF30 and 60i. "
Yes, I have the same charts shot in CF24, CF30 and 60i. I'll have to dig up the footage and check and see if we had to change f-stop when shooting them. I don't believe we did -- I believe light sensitivity stayed the same but I wouldn't guarantee that until checking and verifying.
[Steve Mullen]"Remember how with the DVX100 you can turn ON 2 levels (in the 100A) of blending in the 24A mode. "
You're referring to THICK, MID and THIN. The XL2 also has that option, of LOW or NORM vertical detail. The FX1 didn't have such an option, as I remember. I do believe the Z1 has a different menu option I didn't recognize, I think it was called "allscan"? Anyone with a Z1 who can verify what that does? Maybe that does the same thing?
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 22, 2004 at 12:06:37 pm
ALLSCAN sounds right.
My theory. The FX1 is designed for the consumer world where interlace is the CURRENT standard.
CF25 and CF30 create a progressive like image. If this image at full V rez was sent to an interlace TV all hell would break loose as the camera imaged fine H lines. With 800-lines of V rez the image would look poor on interlace.
So Sony took the precaution of limiting V rez to under 600-lines when in these modes.
In the Z1 -- the user will have a choice of full V rez (840-lines static) in ALLSCAN or NOT-ALLSCAN (THICK/THIN like) under 600-lines.
Clearly if one is making movies or is going to creating an HDTV production -- the Z1 would be your choice. Perhaps one more detail to make the Z1 be the Pro's choice.
Have you measured the V rez difference between THIK and THIN on the VX100?
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 22, 2004 at 4:42:49 pm
[Steve Mullen]"Have you measured the V rez difference between THIK and THIN on the DVX100" In terms of "TV Lines", as measured on an EIA chart, THICK delivers around 380 lines, THIN does 540. A huge difference. And yes, THIN can totally overpower an NTSC television.
JVC developed something they called "high resolution frame mode"... I wonder if it somehow turned off the field blending to give full resolution, and if Sony will do something similar? I never verified what "high resolution frame mode" was...
[Steve Mullen]"Clearly if one is making movies or is going to creating an HDTV production -- the Z1 would be your choice" Of the available choices, yes. But if Panasonic brings forth the rumored 720/24p camera, it might be even more compelling. One thing is certain, this is an amazing six-month span for camcorder buyers -- it's brought forth the XL2, the FX1, the Z1, and maybe (maybe) new products from JVC and Panasonic. Amazing how the landscape has changed so much in so little time!
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by James L. Arthurs on Dec 23, 2004 at 3:39:26 pm
"In the Z1 -- the user will have a choice of full V rez (840-lines static) in ALLSCAN or NOT-ALLSCAN (THICK/THIN like) under 600-lines. "
Steve, I hope you're right about a thick/thin feature control for the Z1, but I think "ALLSCAN" is just Sony's way of saying that the viewfinder/LCD display will underscan and show the full image recorded to tape. Chris Hurd's feature list for the Z1/FX1 mentions "ALLSCAN" and it is placed in the display section of the list...
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 23, 2004 at 4:29:14 pm
Your right.
So it looks like 575 lines for CF30.
Which is about equal to the JVC in 720p30 mode.
So the Sony offers about the same V res and only about 200-lines greater H rez. With true progressive. :)
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by James L. Arthurs on Dec 26, 2004 at 6:24:37 am
Well, if both models are truly only recording one field height as "progressive", that's a bit of a let-down.
I guess in the "glass half full" mode of thinking, at least then you wouldn't have a quality mis-match if you choose to shoot 60i and reduce to fields for slow motion to inter-cut with CF25 material.
I have tried using DV Filmmaker to convert 50i to progressive. Only two clips out there to experiment with, but it definately retains the resolution on static parts of the image. Default settings tend to leave some remains of interlacing in low contrasts areas of the frame. I must do more experiments to see it this can be eliminated...
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 26, 2004 at 1:29:35 pm
That's not really what's happening.
Nevertheless, it seems more and more clear Sony did NOT design any aspect for film making!
In fact, I'm not sure why Sony even bothered with CF 25/30 -- except it could be used for making PAL/NTSC DVDs or high-rez streaming.
CF24 so far seems totally useless.
Nevertheless, for over 15 years 1080i60 and 1080i60 have been used with post production software to make films. That's the way these cameras need to be used.
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Sanjin Jukic on Dec 26, 2004 at 3:09:35 pm
[Steve Mullen]"Nevertheless, for over 15 years 1080i60 and 1080i60 have been used with post production software to make films. That's the way these cameras need to be used. "
But FX1E and Z1 with 50i option are perfectly designed for film use by Sony. 1080i60 for film was always the last option if you are not able to get 1080i50. And 1080i50 is often used and always recommended as the best way in video to film production from the early days of video. Also both options 1080i60 and 1080i50 are on a way. Have a look at DVFilm FX1 page at
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 26, 2004 at 10:22:24 pm
[Sanjin Jukic]"But FX1E and Z1 with 50i option are perfectly designed for film use by Sony. Hate to disagree, but no, I wouldn't say they were "perfectly designed" for film... rather I'd say that they would be the least compromised way of using the FX1E or Z1 to make film-looking video. It's hard to argue that limiting a 1080i camera to 575 lines of resolution, which then needs to be speed-converted down 4%, is a good solution. It's probably a better solution than any other way of using the camera, but it's certainly not a perfect design. 1080/24p would be "perfectly designed" for film use, and that's what JVC's promised. 1280x720/24p would be much better than 1080/50i, and Panasonic will someday deliver that for us.
1080i60 for film was always the last option if you are not able to get 1080i50. And 1080i50 is often used and always recommended as the best way in video to film production from the early days of video." Actually, in Hollywood, 1080/60i and 1080/50i are rarely ever used. 1080/50i has existed a long time, but nobody really bothered to make any movies on high def until Sony introduced the 24P option on the CineAlta. It is the *combination* of 24P and high-def resolution that has revolutionized and energized the shift towards HD production. We've had HD1080/50i and HD1080/60i since 1997, but I can't think of any big-screen productions who have used it to make a movie with.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Sanjin Jukic on Dec 27, 2004 at 9:59:07 am
[Barry Green]"rather I'd say that they would be the least compromised way of using the FX1E or Z1 to make film-looking video. It's hard to argue that limiting a 1080i camera to 575 lines of resolution, which then needs to be speed-converted down 4%, is a good solution. It's probably a better solution than any other way of using the camera, but it's certainly not a perfect design."
I agree FX1E is certainly not a perfect design but is a good one for a cheap HD enabled consumer cam for achieving of a film style look or speed in your picture. Also I do have JVC GR-PD1/625p50 that is a quite OK but you cannot trust this cam because of missing main manual controls. For example auto focus is almost always on even in manual set up and if a light is changed on the focused object than everything is getting out of focus or blurring in your picture. But is somehow a good cam for certain carefully prepared shootings especially because of progressive scan. Have a look how Peter Jackson using it in production of King Kong at
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 26, 2004 at 10:11:45 pm
[Steve Mullen]"Nevertheless, for over 15 years 1080i60 and 1080i60 have been used with post production software to make films. That's the way these cameras need to be used. " If you wanted to make a film with these cameras, yes that'd be the way to go. But actually, I contend that these cameras need to be used for 1080i High-def production. That's where they excel. I don't think they're really suitable for any sort of filmmaking purpose, and that's not what they're designed for (and yes, I full well acknowledge that that's what people *want* to use them for, and *wish* they'd been designed for!) These cameras are interlaced HD cameras, they do a great job at producing interlaced HD, and any other purpose you put them to would be hampering them.
It's like attempting to eat soup with a fork. You could do it, yes, but it'd be so much easier to use a spoon (which is what the spoon is designed for).
I think aspiring filmmakers would be oh so much happier if they just waited a few months to see what JVC and Panasonic deliver on their promise of HD 24P, which would be a much more suitable and usable format for indie filmmaking.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by James L. Arthurs on Dec 26, 2004 at 11:51:21 pm
[Barry Green]"I think aspiring filmmakers would be oh so much happier if they just waited a few months to see what JVC and Panasonic deliver on their promise of HD 24P, which would be a much more suitable and usable format for indie filmmaking."
Here's my take on this; I've owned an XL1 which I sold to buy a DVX-100 which I'm currently selling to put into a Z1.
The Z1 one is a shipping reality at the end of Feb., while any product from JVC or Panasonic will be months away after the announcement months from now at NAB. More likely, these products won't hit the streets until late fall of 2005.
The Z1/FX1 will make the superior wide-screen presentation compared to any camera on the streets (in this price range) even with one field tied behind their back. I've NEVER owned a Sony prosumer camcorder until now, and when the superior offering from either JVC or Panasonic hits the streets I'll gladly sell the Z1 and trade up/into the new product. I am not happy that Sony short-changed us with a restrictive and limited path to a progressive-like image, and will "reward" Sony by jumping ship on the Z1 when anything else is a shipping reality.
Until then, I expect at least six months of time with the Z1 at the top of the food chain.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 27, 2004 at 1:47:54 am
[James L. Arthurs]"The Z1/FX1 will make the superior wide-screen presentation compared to any camera on the streets (in this price range) even with one field tied behind their back."
What a delightful way to phrase it. Yes, even if CF25 offers fundamentally the same vertical resolution as a PAL 25P camera, it will still offer twice the horizontal resolution. And the Z1 does look to overcome most of the "prosumer" restrictions the FX1 had. And, yes, I do agree that any JVC or Panasonic offering will likely be several months after their official announcement, whenever that official announcement may take place.
[James L. Arthurs]"and when the superior offering from either JVC or Panasonic hits the streets I'll gladly sell the Z1 and trade up/into the new product"
I believe that is exactly the right way to view an investment in a camcorder such as this -- as a tool to do the job that you need done, and if another tool comes along that does the job better, or somehow allows you to do your job more efficiently, etc., you retain the right to sell it and upgrade, as opposed to how people seem to get so emotionally attached to their cameras and their particular brands...
"Until then, I expect at least six months of time with the Z1 at the top of the food chain"
I like your attitude. Use the camera, make some money with it, stay current and upgrade until you get what you want... and if nothing else comes out, you'll still have the top dog. Six months is a lot of time to make a lot of money with a camera such as this. Best of luck!
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Gary Taylor on Dec 27, 2004 at 11:11:10 am
Hear Hear!
I am all for celebrating the arrival of the cameras now and "rewarding" Sony when a better camera for your application comes along. Especially when that new camera has features that should have been in the FX1/Z1 in the first place. This brand loyalty trend has gone too far as far as I am concerned and to many vender have been able to segment their products too successfully as a result.
Gary
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 26, 2004 at 10:04:22 pm
[James L. Arthurs]"I guess in the "glass half full" mode of thinking, at least then you wouldn't have a quality mis-match if you choose to shoot 60i and reduce to fields for slow motion to inter-cut with CF25 material. "
Well, no, there still will be a little bit of a mismatch, but not nearly as much as on a DVX or XL2 where you're mixing progressive with interlaced. On the FX1 it appears that you're getting a raw field's worth of resolution, not a blended field, so you're looking at 575 lines (vs. the about 400 if you'd shot 60i and slowed that down to 50%). CF30 (and presumably CF25) are going to be a lot higher-res than doing a de-interlace in post.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by James L. Arthurs on Dec 26, 2004 at 11:38:17 pm
[Barry Green]"Well, no, there still will be a little bit of a mismatch, but not nearly as much as on a DVX or XL2 where you're mixing progressive with interlaced. On the FX1 it appears that you're getting a raw field's worth of resolution, not a blended field, so you're looking at 575 lines (vs. the about 400 if you'd shot 60i and slowed that down to 50%). CF30 (and presumably CF25) are going to be a lot higher-res than doing a de-interlace in post."
Barry, then I'm completely confused. Follow me and tell me where I've gone wrong;
Your assumption has been that the CF mode in NOT taking advantage of two fields in any way, but is just a single field, correct? If so, it would have a potential vertical resolution of 540 true lines (not 575). If you shoot 60i, then separate into fields, lift the second field up one scan line and play out the sequence as individual frames then you have a slow-mo stream of images 1/60th sec apart with 540 true vertical scan lines each (not 400).
If you're measuring greater than 540 lines of resolution in a CF mode, then the CF mode HAS to be taking advantage of extra info from the second field. If that's the case, then WHY would it only add up to 575 lines? Now of course this is all "potential resolution", and measured results would be less, but if CF limits vertical resolution to 1 field how do you get a number greater than one field?
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 27, 2004 at 1:34:03 am
[James L. Arthurs]"Your assumption has been that the CF mode in NOT taking advantage of two fields in any way, but is just a single field, correct? If so, it would have a potential vertical resolution of 540 true lines (not 575). I measured 575 lines. However, I believe that it's close enough to the "margin of error" that 540 may be the actual correct number, which would be one raw field. However, that's assuming a field that's unprocessed/blended/flicker-reduced, which doesn't happen when shooting 60i. In 60i, you get about 775 lines of resolution, which is close enough to Steve Mullen's theorizing about it delivering a max. of 820 lines that I figure it's within the margin of error as well, and just round it off to 400 lines per field.
If you shoot 60i, then separate into fields, lift the second field up one scan line and play out the sequence as individual frames then you have a slow-mo stream of images 1/60th sec apart with 540 true vertical scan lines each (not 400). " Well, no. If shooting interlaced gave you 540 lines per field, then you'd be correct, but if you read Steve's article you'll see that you only really get somewhere around 390 to 410 lines per field (because of blending/blurring done to reduce interlace flicker).
So what I'm saying is, CF25 is BETTER than shooting 50i and de-interlacing, because 50i->de-interlace will give you about 400 lines of resolution, whereas shooting CF25 gives what appear to be 540 to 575 lines. My GUESS is that how they're accomplishing it is by giving us just one raw unblended/unprocessed field, simply because the observed 575 is so close to the theoretical 540 that it just seems likely. However, we don't know for sure how Sony is delivering CF25. It just seems reasonable to say that it'd be a single field. But that single field is higher-res than a de-interlaced field from the 60i stream would be.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 27, 2004 at 4:50:15 pm
"So what I'm saying is, CF25 is BETTER than shooting 50i and de-interlacing, because 50i->de-interlace will give you about 400 lines of resolution, whereas shooting CF25 gives what appear to be 540 to 575 lines."
I strongly disagree. A well designed software deinterlacer will, for static images and static objects deliver all 810-lines of effective V rez. Only moving objects will suffer a loss in rez. And, since the object is moving and subject to motion blur -- you'll not notice the loss in V rez!
Thus shooting interlace is far better than CF25.
While 1080i50 is a simple way to handle video -- handling audio is a nightmare.
If one is in the USA, then shooting 1080i60 is best. The software to deinterlace and generate 24fps is cheap ($100) and once in 24fps it's trival to edit video AND audio in sync.
There are dozens of 35mm films shot with 60i starting with Frank Zapas' "200 motels." All DV, Beta SP, films shot in the USA have been done with 60i. Think Hoop Dreams, Timecode, Julian Donkey Boy, etc.
The only reason PAL was popular was the extra rez. That's not an issue in HD. Plus, the horrible 50Hz refresh rate makes using any 50Hz format totally dated. Plus, try to find a 1080i50 HDTV. I read the Euro mags and they are, if you consider Japan, nearly 20 years behind the NTSC world.
And if you do find an HD panel or projector, they also display 720p60 and 1080i60.
I find it vrey odd that anyone in Region 50 would be recommending going that way. I'm in India and all the TVs show both PAL and NTSC. Moreover, Region 50 products usually cost more. Makes no sense to me. :)
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 27, 2004 at 4:50:17 pm
"So what I'm saying is, CF25 is BETTER than shooting 50i and de-interlacing, because 50i->de-interlace will give you about 400 lines of resolution, whereas shooting CF25 gives what appear to be 540 to 575 lines."
I strongly disagree. A well designed software deinterlacer will, for static images and static objects deliver all 810-lines of effective V rez. Only moving objects will suffer a loss in rez. And, since the object is moving and subject to motion blur -- you'll not notice the loss in V rez!
Thus shooting interlace is far better than CF25.
While 1080i50 is a simple way to handle video -- handling audio is a nightmare.
If one is in the USA, then shooting 1080i60 is best. The software to deinterlace and generate 24fps is cheap ($100) and once in 24fps it's trival to edit video AND audio in sync.
There are dozens of 35mm films shot with 60i starting with Frank Zapas' "200 motels." All DV, Beta SP, films shot in the USA have been done with 60i. Think Hoop Dreams, Timecode, Julian Donkey Boy, etc.
The only reason PAL was popular was the extra rez. That's not an issue in HD. Plus, the horrible 50Hz refresh rate makes using any 50Hz format totally dated. Plus, try to find a 1080i50 HDTV. I read the Euro mags and they are, if you consider Japan, nearly 20 years behind the NTSC world.
And if you do find an HD panel or projector, they also display 720p60 and 1080i60.
I find it vrey odd that anyone in Region 50 would be recommending going that way. I'm in India and all the TVs show both PAL and NTSC. Moreover, Region 50 products usually cost more. Makes no sense to me. :)
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 27, 2004 at 5:02:24 pm
"I measured 575 lines. However, I believe that it's close enough to the "margin of error" that 540 may be the actual correct number, which would be one raw field. However, that's assuming a field that's unprocessed/blended/flicker-reduced, which doesn't happen when shooting 60i."
This is the question!
Either:
1) the CCDs are switched to a mode that turns OFF row summation thereby yielding 1 field of 540-lines.
2) The camera shoots regular 50i/60i and then smart deinterlaces this to yield 810 to 540 lines which is then filtered by 70% to 575 lines. Why 70% -- turns out that is exactly the amount the DVX100 filters to get to THICK.
Looking at Sony's spec sheet on CCDs it is NOT possible to turn off ROW SUMMATION. In short, I don't think the CCD can generate a full 540 line field -- only 405 lines. That's way too low to measure 575!
DID ANYONE MEASURE CF24?
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Sanjin Jukic on Dec 27, 2004 at 5:29:45 pm
I am a bit confused with all these counting of the lines etc. Only what I surely know is that FX1E when I capture delivering the picture size 1440x810 pixels that V could be actually 810 lines, and that JVC GR-PD1E progressive cam in capturing deliver 1024x576 px or V=576 line, specs wrote 625 line that is even better. Finally to say that Panavision Genesis CMOS digital camera shooting from 1 to 50 frames per second and it is Made in USA. So 60i is just left over from NTSC and does not mean that is the best speed solution for digital cameras. What can I say more? 50 FPS or 25 FPS remains as the best in interlaced or progressive world.
Have a look at Panavision Genesis site
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Barry Green on Dec 27, 2004 at 7:16:54 pm
[Steve Mullen]"DID ANYONE MEASURE CF24? "
Yes, I've got the same chart in CF24. It's difficult to measure the vertical resolution, because you can't just extract a still -- CF24 "pulses" with a rhythmic jitter/flicker that changes the vertical res number! Besides the erratic/uneven motion rendition, it is this pulsing and resolution change within frames that makes CF24 unusable, IMHO.
However, viewing stills from a group of five frames, it looks like the lowest res is 425, and the highest res is 575. It looks like there are two frames of 425, two frames of 475, then one frame of 575 in each group of five.
Here's an uncompressed .AVI extraction of a few seconds, so you can see the pulsing for yourself. It's a 176x120 extraction, about 10mb large:
http://www.icexpo.com/FX1-CF24-Res.avi (be sure to right-click and "save as...")
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 27, 2004 at 8:19:05 pm
"However, viewing stills from a group of five frames, it looks like the lowest res is 425, and the highest res is 575. It looks like there are two frames of 425, two frames of 475, then one frame of 575 in each group of five."
If you look at my CF24 diagram, the first frame should have max V rez. Two of the frames should be judder frames with the content of two different fields -- the 475 frames.
And 2 that have simple interpolation -- the 425 frames.
OK -- THAT'S MY INTERPRETATION of your data. :)
Interesting that once again, the maximum rez is about 575 which I explain by Sony imposing a low-pass filter on ALL the CF modes to prevent line flicker and twitter when "progressive" is displayed on interlace TVs.
HOWEVER, are you inputing via firewire? If not, the filter could on analog output.
Makes the CF modes OK for interlace display. But damn near useless for film making when compared to the great results from software deinterlacers.
Remember that both CF25 and CF30 will have exactly the same EYE STROBING problem as 720p30. Why worry about that when you have the option of 50i and 60i?
And, Sanjin if 60i comes from NTSC then 50i comes from the old fashioned inability of Britsh AC generarors to crank out 60Hz. There is no excuse to be living with 50Hz flicker in the 21st Century.
Moreover, 100Hz Euro TVs suffer from the same eye strobing artifacts as does 720p30 in the JVC. Presenting the same frame twice is not a good way of raising the flicker fusion frequency. (One reason film projectors limit light output and yet you still see flicker in bright skys.) That's why computer monitors run at 85Hz to over 100Hz.
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by sanjin jukic on Dec 27, 2004 at 8:44:38 pm
[Steve Mullen]"And, Sanjin if 60i comes from NTSC then 50i comes from the old fashioned inability of Britsh AC generarors to crank out 60Hz. There is no excuse to be living with 50Hz flicker in the 21st Century. "
Steve,
I know that I am an amateur for all that kind of scientific explanation about how these HD cameras work but I have one question about Panavision Genesis. It is written the following: "Features: 1 to 50 frames per second".
So how is possible that this high-end HD camera can shoot maximum 50 FPS only? If I compare that fact with your story about 60 Hz I could conclude that something is wrong about ideal shooting with of 60 Hz interlaced or progressive doesn't matter. Or maybe something is wrong with my opinion about 50 Hz or with your's about 60 Hz or Panavision's about to get maximum FPS at 50 Hz. Where is the truth? what should be the best? Or Panavision could not get 60 Hz or more with the CMOS they are using? Or parameters are different between CCD and CMOS? Or...? The numbers are sometimes confusing, isn't it?
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Steve Mullen on Dec 29, 2004 at 1:58:59 pm
This camera is for shooting 24fps. 50fps is just for over-cranking.
Steve Mullen
Digital Video Consulting
www.mindspring.com/~d-v-c
===
HDpartner Prime -- Real-time 1080i and 720p HDV Editing with FCP HD
GR-HD1 and JY-HD10 HDV Shooting Guide
AG-DVX100 Shooting Guide
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by Sanjin Jukic on Dec 29, 2004 at 2:40:57 pm
And I am quite satisfied with a picture quality of FX1E 25 FPS converted to 24 FPS solution till better camera appears for the same price. From Panasonic, Canon or who else...and 1080p24 for that price is just a dream.
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by simkiss on Jan 7, 2005 at 4:31:38 pm
Cameras in reference: Sony HDR-FX1 and HVR-Z1U
QUESTION GROUP 1:
If you film at 60i, your datastream is 25 Mbps on the Z1/FX1. Likewise, if you film at 50i (Z1 only), the datastream is still 25 Mbps, isn't it?
Since HD resolution for 50i/25p and 60i/30p is the same...
(unlike NTSC's 525 lines @ 29.97fps vs. PAL's 625 lines @ 25fps for example making the datastream bandwidth equivalent between the two...both 50i and 60i record at 1440x1080 and both output 1920x1080i)
...doesn't that mean that at 50i there is 20% less compression of the video since there are 20% fewer frames at the same resolution??
QUESTION GROUP 2:
I will use the Z1 to record for blow-up to film. Has anyone seen this done yet (with an FX1 since the Z1's not out yet)?
Since my output will eventually be 24fps, taking into account question group 1 above, should I record in 60i or 50i? I saw the resolution loss for CineFrame 24/25/30 with the links above--pretty bad. I guess I'm using software-based conversion from i to p. Who's the best without requiring indentured servitude to pay for it?
Re: Cineframe 30/Cineframe 25 resolution loss vs. 60i by simkiss on Jan 11, 2005 at 5:33:00 pm
I've got the answer. 50i uses the full 25Mbps as well! It's confirmed. That means 20% less compression vs. 60i since they're both the same resolution and both 25Mbps. Woohooooo!!!!!
Can someone perform a test like this one ( http://www.icexpo.com/FX1-60i-vs-CF30.BMP ) at 50i with the FX-1E or the Z1 so we can see how many lines we get? (I don't know how it's done or have the stuff to run the test)
