Any reviews with hard facts?
 

Fx1 has been in stores for more than a month now and demo pieces of z1 are travelling around the globe.
Is there still any review published about these cams, that would have been done with scientific methods in lab environment by professional people?
Just the basic things like resolution, latitude, lens distortions with diffrent aperture, shutter speeds and cf-modes.

Comparison with other cams would be useful as well as with uncompressed (analog component hd's bitdepth?) and compressed output (tape & fw).

Especially resolution with slow shutter & cf interests me, because I was speculating before, that these ccd's might be progressive after all and if they are then Reel Stream's 14/12bit 4:4:4 -hack would sound amazingly interesting.

Still about the ccd's, this is what Steve Mullen says about them:
"How difficult would it be for the Pro version of the FX1 to support 24p video? Perhaps not difficult at all — if the CCDs' frame rate can be set to 24fps instead of 30fps. CCDs are inherently progressive; thus, every CCD readout yields 1080 lines. For interlaced scanning, “row pair summation” creates a 540-line field from each scan. However, were summation disabled and every other readout discarded, each 1/24 of a second, all 1080 lines from the CCDs would become a progressive frame stored in a buffer. From the buffer, alternating “fields” of 540 odd and 540 even lines would become available as needed. (The fields, of course, are not interlaced.) This mode of operation is very similar to the 1080/PsF24 recording used by Sony CineAlta cameras."

What I'd like to see is a test that shows a comparison of the DV mode of the camera, the HDV mode, and the same thing that was shot in the HDV mode and then transferred to DV via firewire.

I can do that soon, DV and HDV.

heath

That would be cool. What I'm really interested in is what the video would look like if you shoot something HDV, then transfer it to DV, as compared to shooting the same thing DV. Just curious if there is a difference and what it would be...wondering if the higher res of HDV would make a downres image look any different, better or worse, than originally shooting in the DV mode.

Anything that Heath comes up with, we'll run his reports here in a big way.

Excellent. This is probably a dumb thought, but I was thinking about how we shoot 16mm and then have it transferred to Digibeta, usually, and then to DVCAM for editing, and that looks better than anything shot original DVCAM; just as HD downrezzed to DVCAM looks better. So I was wondering if the higher res of HDV, when transferred down to DV/DVCAM, might perhaps look better than the same shot originated by the same camera in DV/DVCAM.

I have a clip of Kaku downrezzed to 30P mpeg NTSC DVD which looks stunning. If Chris can host I can up (~50meg)

I was thinking of doing that exact thing. Shoot HDV, change tapes, shoot DV, and then use the old Canon ZR10 to shoot the same thing. I am taking the little camborder along as a rewinder / tape retensoner, anyway.

I could post stills of the results I suppose. Although I can't imagine any real difference between HDV exported as DV, and shooting DV to export as DV. It would be interesting though. What if DV exported as DV looked better than HDV exported as DV? What would that do to our workflow?

I've been running all sorts of tests on FX1 vs DVX vs XL2, and most recently we've been focusing on the aspect of "how does it look on DVD"?

On a high-def TV, obviously there's no contest, the FX1 is high-def. But for the next several years, content is going to still be distributed on DVD, so how does the FX1 compare to straight DV cameras when producing content for release on DVD?

There's a huge thread running on another forum with all the results so far, but I don't necessarily want to point anyone out of the dvinfo.net family. If I could post images in-line in a thread here I'd do so, but there are so many pictures that it would be awkward to constantly refer someone out of the thread to a picture and then back.

In quick summation, my conclusion is that for DVD release, the DVX and XL2 produce a superior picture on DVD than downrezzed FX1 does, whether the FX1 is downrezzed in-camera (outputting DV from the firewire port) or downrezzed in post (using Vegas 5.0, "Best" video quality for bicubic resizing, outputting from the straight camera original .m2t transport stream). It's not a night-and-day difference, and other than the sensitivity issue it could be argued that all three cameras are producing basically the same resolution, with the FX1 delivering a slightly, tiny bit softer image at DVD resolution, but certainly all are within spitting distance of each other. Where the FX1 takes a step back, for SD DVD production, is that it doesn't offer all the other features of the XL2 and DVX, such as 24P or 30P, XLR audio, and it's the least-light-sensitive of the group.

The FX1 makes great high-def images, that's for sure. But if you're looking to improve the video quality of your DVD's, the FX1 will not help you, as compared to the DVX and XL2.

Barry, technically I believe the opposite is true. If we ignore lens and sensetivity characters and use the common mode 30p for comparison; the best SD source and best HD source downscaled to SD will have equal luma resolution 720x480. Note: HD's oversampling can have an edge in luma, yet good SD cameras oversample these days anyway (e.g. XL2), so they're equal. Chroma is a different matter, DV NTSC has a maximum chroma res of 180x480, whereas the FX1 has a maximum chroma res of 720x540. Given that DVDs are 4:2:0 with a maximum chroma resolution of 360x240, only the HDV source can achieve the full chroma resolution of a DVD. DV converted to DVD gives you a maximum chroma resolution of only 180x240. For this reason I would not recommend using the HDV->DV path, as result are better when editing HD then doing a down res to DVD on output.

Or you is edit in DV50 mode like with DVCPRO 50 codec from Apple. It will scale down FX1 footage to 4:2:2, good enough for real-time editing and final distributing on DVD, without need to edit in HDV. Of course, edit in HDV if you thinks you will need a HD master later on.


FOr this very reason, I think it is prudent to use FX1 as it is better than any SD DV camera for the color resolutions alone.

Is chroma res of NTSC DV, 180x480? It is 180x480 I think. Remember only 1 chroma sample for every 4 luma samples, so if luma is 720, divide by 4 and you is getting you correct number of 180x480. Very same indeed.

Like in my last posting, if you downrezz FX1 footage to 100% photoJpeg or MJPEG, you WILL GET 4:4:4 CHROMA!!! For this reasons alone, the FX1 is MUCH BETTER than XL2 or DVX100. Then also the PAL version can shoot 50i, convert it to 24P NTSC and it blows away DVX100 footage. I can guarantees that much.

Has anybody ever showed anykind of resolution chart to these hdv cameras?
Another thing that interests me is chromatic aberration.
With my fast hands on test aperture full open and max tele, I saw quite lot of it.
So comparison eg with xl2's "not hd" lense would be very nice...

<<<-- Originally posted by Davi Dortas : Is not chroma res of NTSC DV, 180x120? It is not 180x480 I dont think. Remember only 1 chroma sample for every 4 luma samples, so if luma is 720x480, divide by 4 and you is getting my correct number of 180x120. Very bad indeed.-->>>

You have you math a little wrong, that would end up with one chroma for every 16 luma. :) 4:1:1 subsamples luma horizontally only. Chroma res for NTSC DV is 180x480.

David, I know what you're saying, and theoretically it sounds good. But in practice, it's not happening. The XL2 and the DVX just plain look better, on DVD, than the FX1 does.

I'm taking footage shot side-by-side, same identical frames in time, from each camera, and split-screening it and making a DVD of it. I'm using the FX1 original HDV source, rather than using a DV proxy of it, specifically for what you said. If you convert FX1 to DV first, you'll force it through a downrez and a color space conversion, and then when it goes to DVD it'll go through another compression cycle and another color space conversion.

The way I'm doing it, it goes through one downrez and no color space conversions.

But the end result is, the DVX and XL2 footage looks slightly sharper and more detailed with better color. The downrezzed FX1 footage is close, yes, but it's not in any observable way "better".

I agree, and that's the way I've been doing it.

<< If I could post images in-line in a thread here I'd do so, but there are so many pictures that it would be awkward to constantly refer someone out of the thread to a picture and then back >>

Actually Barry if you wanted to share some of those images, I could put 'em on a page (or pages) on the content side of this site (at www.dvinfo.net/articles, which would open in a separate window -- no need to send someone out of a thread. Just wanted to make that option available to you.

Barry,

Therefore sounds like a scaler issue as the HDV will have more chroma detail and equal luma detail. There are many factors to consider when doing the scale. Are you doing an interlace mode scale (even if you shoot 30p?) This will greatly limit for scaling quality. Also are we talking real detail or sharpness? The DV sources are slightly sharpened, whereas the post sharping is lost when you down-scale. You may need to apply an unsharpen mask to your scaled footage to match the processing path that the SD footage went thru in camera.

35mm downcoverted to DV still looks better than DV, so I can safely assume HDV converted to DV will look great and noticable vs. native DV.

heath

You can't shoot 30p on the FX1, as of course the camera is only capable of 60i. I've downrezzed CF24, CF30 and 60i modes.

I'm talking about observing the footage of the three cameras simultaneously, and seeing the FX1 footage looking "soft". All three cameras were set to mid detail level (the Sony defaults to +12, I set it down to 8 which is the middle...)

I'm going to have to take Chris up on his offer to host pictures. I went through some convoluted exercises to get to the bottom of it, and have some pictures to show why. Bottom line is, downrezzed FX1 does look a little bit softer than straight DVX and XL2. I've heard people proclaiming that it looks so much better, but I've been unable to find any sort of supporting evidence to back that up. I understand the theories of why it is supposed to be, and I know that people want it to be, but I'm not really all that concerned about what "should be", only what *is*. I'll gather the pics and stuff and have Chris put up an article, and then you guys can go through and see what I'm seeing, and if there's something I'm doing wrong, let's hear it.

Ah, but Heath, you know what they say about assuming, of course. 35mm to DV is such a completely different animal from downrezzing a digital HD source, that it's entirely not possible to compare them. 35mm is scanned by an optical CCD, whereas HDV is a digital image that's being digitally resized. Completely, thoroughly, totally different animal.

If you want to explore the 35mm analogy, the only way to make it apply would be to say: would 35mm telecine'd to HDV and then downrezzed to DVD look better than 35mm telecine'd to standard-def and then put on DVD? And the answer, I strongly suspect, is no.

I know exactly why you said what you said, and it does seem like a safe assumption, but that's why I run tests: to find out what *is*, rather than just accept what's *supposed to be* because frequently they are not the same thing. It does look good, it looks almost the same (resolution-wise) as native DV from the XL2 and DVX. But not better. People have gone so far as to say that downrezzed FX1 footage "blows away" these other cameras on DVD. I'm saying that's completely false, and downrezzed FX1 looks *almost* as good as the other cameras. But not better.

so Barry, that would mean if my film is to be DVD release then I am better off stick with DVX100a and dont sell it to replace FX1/Z1 ...as I am sure it will be a DVD release as the previous one shot on DVX . Is that safe to assume at this moment....sorry not assume but from the tests you have conducted?


Thanks

<<<-- Originally posted by David Newman : Barry, Therefore sounds like a scaler issue as the HDV will have more chroma detail and equal luma detail. -->>>

Could it be mpeg2 that blurs the picture?
There's usuallly some amount of macroblock's edge blurring and details are lost within macroblocks due to DCT.

Barry did you also do the tests with FX1 at its default of +12 since this is the way it will be used most of the time on PP1, with no PP set to off ( what ever that does to the settings!!) and at DV to see if there was a real difference. Have you also used the Cineform wavelet intermediate as a transfer to DVD?

Ron Evans

Barry,

My HD10 shot in HD mode and down-converted to DV looked better than when I shot on DV mode.

heath

<<<-- Originally posted by Ron Evans : Barry did you also do the tests with FX1 at its default of +12 since this is the way it will be used most of the time on PP1, with no PP set to off ( what ever that does to the settings!!) and at DV to see if there was a real difference.
Ron Evans -->>>

I too wonder that - why cripple the camera then complain about it's relative lack of sharpness?

Also- another thing to consider- the HDV footage will hold up much better during multiple render passes, effects, etc. Any artifacts introduced whill be hidden during the HDV > DVD down res. While the artifacts on the SD footage will remain. I still find it hard to believe the FX-1 footage doesn't look as good- I've done a good bit of screen grabs and when I resize the images from HDV resolution to DV resolution, the resulting images are always much crisper than a DV screengrab. Now I realize I'm comparing the FX-1 to the FX-1 but the difference in images is HUGE.

I'm curious to hear your exact workflow, including settings, etc.

- I'm also curious to compare the HDV 4:3 center cut output of the Z1 vs the same footage in DV.

Also- the FX-1 default is +11 not +12.

Video files
 

Barry could you post both raw DV (from DVX100/a) and FX1 m2t files of the same scene (about 5 seconds or whatever suits you), i would like to test the downres for my self thanks Barry!

I agree, source files would be handy as the scaling can be done to achieve the desired results.

Barry,
Film telecined (or scanned) to HD (or 2k -- 10% more than HD) then down res'd to SD also looks great. It is a standard industry workflow. In a digital intermiate workflow, color correction is often done on 1k proxy files (around 960x540), for film out the 2k (or greater) is processed, but for the DVD these 1k proxies of often used (are looks perfect.) Oversampling works, all that signal processing theory can't be wrong -- only implementations can be.

What I'm saying is, for DVD release, the FX1 offers no advantage over the other cameras. So you wouldn't want to replace a DVX with an FX1 just to get a better-looking DVD, because it won't look better. In interlaced mode it'll look about the same, maybe a tiny bit softer, but the DVX will give you two stops more sensitivity, 24P, more picture control, more manual control, better audio, etc.

The FX1's advantage (and it is a big advantage) is for HD content creation. But if you're not releasing on HD, that advantage is nullified, so you're left comparing the cameras at SD resolution. And in any direct comparison between them at SD, the FX1 doesn't do as well as the XL2 or DVX. (except when it comes to picture noise, where the FX1 is quite a bit cleaner than the DVX in the shadows).

Certainly it does. However, keep in mind that I'm comparing all three cameras split-screened into the same frame, so any MPEG artifacting/blurring would be applied to all three images equally.

I did not. I tried to neutralize the effect of the various settings. I mean, we could have also tested with the DVX at +7 and the FX1 at +15, and the DVX at -7 and the FX1 at 0, and every combination in between, but there's only so much time in the day and so many ways you can test something. So I set 'em at neutral (or what I expect would be neutral).

Did not use the Cineform, I wanted to go from the original source as that would be the highest-quality and most accurate way to go.

Cripple? That's an odd choice of words. You could also say the XL2 and DVX were "crippled" because I didn't artificially introduce more sharpness/detail/edge enhancement into their picture by cranking up the settings, couldn't you?

The attempt was to get a natural-looking picture, not one that was excessively artificially sharpened.

Re: Video files
 

Yes, I always intended to do that. Only problem is, all the files I captured are way huge, and I have no way of outputting FX1 m2t files. I'll have to re-install Cineform on a system and capture little five-second snippets. I'd love to see some outside testing to verify or disprove what I'm finding.

Yes, it does look great. And downrezzed FX1 looks great too. Just not better than the competing cameras shooting native.

I'd heard some reports that said downrezzed FX1 "blows away" cameras like the DSR500, DVX, XL2, any other camera... all I'm saying is that in all the tests I've run, that's simply not true. It looks, at best, about the same. Maybe a little softer. But not "better" by any way that I can see.

<<<-- Originally posted by Barry Green : Certainly it does. However, keep in mind that I'm comparing all three cameras split-screened into the same frame, so any MPEG artifacting/blurring would be applied to all three images equally. -->>>
I meant that those mpg2 artifacts blur also the original hdv material.
Just trying to find reason why hdv doesn't have more details than dv.

But before somebody takes good objective measurements from fx1/z1's real resolution, this is all just speculation.

Is there anybody professional who has resolution charts, knows how to use them and owns a camera and after all would be interested to know how good is it?

Or is it so that new camera owners are scared to find out that interlaced hdv of fx1/z1 doesn't have better resolution than progressive dv from dvx/xl2?
This is possible if you take out kell factor from hdv's resolution and then think how much details it's heavy mp2 compression removes in dct.
Another thing that might give the impression that hdv has much more resolution might be that hdv is usually watched from hd display.
Dv is usually watched from display of with connections that can't show the full horizontal resolution or progressive picture.
Isn't even analog sd component signal's luma limited to below 5 MHz?

<<<-- Originally posted by Barry Green : I did not. I tried to neutralize the effect of the various settings. I mean, we could have also tested with the DVX at +7 and the FX1 at +15, and the DVX at -7 and the FX1 at 0, and every combination in between, but there's only so much time in the day and so many ways you can test something. So I set 'em at neutral (or what I expect would be neutral). -->>>

Of course nobody has time to test every combination, but testing the min and max settings would tell all of us what the limits are.
After that everybody can imagine what's between.
So now we know what camera give's at default settings, not what camera can give.

Toke,

I am currently scheduled to do a shoot with the FX1 for Heath McKnight this week. As part of my initial tests, I plan on putting my Putora 7A9 chart in front of both the FX1 and the DVX100A. I will be posting my thoughts on my experience with the camera and my opinion on its performance on this board as soon as I get back to NY. And it's likely that Chris will eventually post my review in article form on the HDV Info page in the Articles section as was done with my reviews of the HD10.

It will be up to Heath and his colleagues to post any footage or still frames for public viewing. Though I'm sure it won't be too much hassle to post some comparitive stills of the FX1 and DVX100A in front of the 7A9 chart.

Jon

Barry

My understanding of the resolution issue you are observing is as follows:

When downconverting 1080 footage to 480 using low order interpolation methods you are basically passing the footage through a low pass shelf filter. In question you are averaging every 9 pixels to 4 pixels vertically. A frequency plot of such a filter will show that there is a reduction in response some way before the cutoff frequency (which is the 480 scan line frequency in this case) is reached which is the reason for the slight image softening. The cure is to use high order interpolation (fifth and above, not bicubic) which will notch boost the response near the scan line frequency. Rather than go to all that trouble a simple edge sharpener will be just as good. Without such a cure you are comparing unsharpened footage with sharpened footage, since your DV cameras are at default sharpness - which means they are sharpened, and because, any sharpness in the original HDV source is stripped out by the filter action of the downconversion.

One must be careful having sharp footage in that the higher frequencies present which are attributed to sharpness may be close to the scan line frequency which will give rise to an unsightly pnenomenon known as twitter, which can only be cured by softening the footage.

I can list a number of advantages of downconverting HDV>DVD over and above DV>DVD

- reduction of noise by > 50%
- removal of the halo effect
- reduction of artifacts > 50%
- improvement of any slight de-focusing faults
- improved colour space going into the mpeg compression for DVD
- choice of whether to have interlace or progressive DVD
- better framing from cropping if desired

this to my mind is better , not worse or equal


At the risk of repeating myself (not to you specifically) I have a DVD encoded sample clip of Kaku which I can upload to anyone who cares to host it.

Whoa -- let's hold up here. HDV has way more detail than DV. Way, way more. If you watch the original HDV source on a 1080i television, the difference is startling.

The subject that we've been discussing is, how does that HDV hold up when it's been downrezzed to DVD. That's where I'm saying it's basically equal, and HDV holds no advantage over comparably-shot DV from a comparable camera.

But to say that HDV doesn't have more detail than DV is to totally miss the point. Either I misunderstood you, or you misunderstood me, so let's be clear about that. In HDV mode the resolution is dramatically superior to any DV camera.

It's just that when you down-rez it down to DVD, the HDV image doesn't retain any advantage over comparably-shot DV.

Jon:

<< it's likely that Chris will eventually post my review in article form on the HDV Info page in the Articles section >>

I'm hoping to replace that word eventually with very soon. Thanks,

John, I agree in theory with every advantage you just listed for HDV->DVD. Especially about noise, as the downsampled FX1 picture is very clean (although I don't believe it's cleaner than the native XL2 signal).

However, in practice, I'm just not seeing the results.

Chris has stored the pictures on his site, so here's the breakdown of how I arrived at my conclusions:

I've got hours of footage we're trying to compile into a meaningful side-by-side. I'm split-screening XL2/DVX/FX1 footage showing the EXACT SAME SETUPS, EXACT-SAME LIGHTING CONDITIONS, EXACT SAME CAMERA MOVEMENTS, etc... only difference is the cameras themselves.

And I'm giving the FX1 the benefit of the doubt by making the DVD source from the original HDV footage, not a DV down-rez, because by using DV down-rez, you convert the original 4:2:0 MPEG-2 down to 4:1:1 DV, and then again to 4:2:0 MPEG-2 for the DVD authoring. Whereas with what I'm doing, it goes from 4:2:0 MPEG-2 to 4:2:0 MPEG-2, no DV downconversion.

I'll post screen grabs, screen grabs are a nearly pointless way to evaluate video quality -- it's video, not stills, and so much of the experience relates to motion artifacts and motion handling and perceived resolution and blah blah blah. So people can ask for stills, but other than illustrating specific items (such as, see how much brighter this camera is than that one) it's not a good measure of examining video performance at all.

I've put this stuff on DVD and viewed it on a computer screen, on a conventional TV and then also projecting on a 10' screen using component inputs, and standing like 2' away. There is no circumstance that I've found where the FX1 footage looks any sharper than the DVX or XL2 footage, when rendered to DVD. Even in interlaced mode, they look (at best) the same, sharpness-wise, or at least within spitting distance of each other. On CF24 footage vs. progressive, the DVX and XL2 hold quite an advantage, especially when the FX1 had Cinematone gamma on -- all the dark sections get so crushed down on the FX1 that you can't see any shadow detail vs. the DVX, which shows everything.

Yes the signal on the FX1 is cleaner with deeper blacks. And I don't know but I suspect that under a more controlled shooting environment, cranking up coring and lowering the master pedestal, it may be possible to get the DVX to match those attributes of the FX1. Even if not, the DVX is still two stops faster, with what appears to be more latitude, and genuine 24fps movement instead of the fake CF24.

If I'm wrong, I'm glad to be proven wrong -- I don't like to continue being wrong. But I just don't see it. In HDV mode on a high-def monitor, yes the FX1 does a great job (on still shots, a less great job on moving shots, but it's still high-def). But when put on DVD... the tables shift, and the XL2/DVX just look every bit as good and usually better.
(continued)