Breaking News: Sony HVR V1U (USA model) progressive mode is also affected!

[digiteye]

Now people are beginning to see the problems with progressive mode even on the V1U. The artefacts are the same as reported for 25P except for some reason they are slightly reduced.

Sample clips and .jpg-s available.

http://www.dvinfo.net/conf/showthread.php?t=84373&page=4

Check also other pages of the same thread.

Direct link to comparison shots:

www.geekstudios.com/demos

:shocked:

[Sean Michael]

About a month ago, I was asking my local camera shop about the V1U. A salesman there warned me against buying an early release Sony camera. He stated that Sony cams have a history of issues (audio bugs, etc.) in the early going. I didn't think much of his comments at the time...but the current V1E/V1U situation makes me wonder. It's a shame because I've been strongly considering this camera. Now I have serious doubts about it. :undecided

[philnerd]

About a month ago, I was asking my local camera shop about the V1U. A salesman there warned me against buying an early release Sony camera. He stated that Sony cams have a history of issues (audio bugs, etc.) in the early going. I didn't think much of his comments at the time...but the current V1E/V1U situation makes me wonder. It's a shame because I've been strongly considering this camera. Now I have serious doubts about it. :undecided

Well speak of the devil, I just commented about those new samples in another thread here...

Anyway, if progressive is important to you, I'd recommend against this cam. I know Steve Mullen keeps posting samples of nice frame grabs, but unfortunately the other samples coming from end users now cannot be ignored. If shooting high quality progressive footage sometimes is good enough, this cam might be OK. Otherwise the Canon XH A1 is the best bet in this price range.

Bummer, an otherwise really neat camcorder.

[John Trent]

I'll understand if this post is removed. I don't know if I'm doing something inappropriate here or not. Sorry, if I am. As always, I'm only interested in finding out how things work, or don't work. This was posted at sonyhdvinfo and it was said that one V1E owner was "advised" the following:

"The V1's hardware simply cannot cope with the additional bandwidth requirements of encoding progressive material. This causes the encoder to run inefficiently and herefore create a lot of macro blocking. The digital noise reduction circuit then tries to smooth out these macro blocks destroying fine detail causing the oil paint effect. There will be no more fixes. It will require a new encoder and that is not going to happen until the next camera. The issue is closed"

It wasn't made clear who "advised" the owner with this information. I don't see why the U.S. models would be any different if this IS the case. I like Sony and I'm not trying to trash them but I'd like to know from the tech guys here if this sounds like what's going on. I've always thought 24P and HDV's GOP might not play nice together. The JVC HD100 with it's 6 frame GOP was "sticking" maybe this is what happens with a 12 (PAL) or 15 (NTSC) GOP? And maybe this is why Canon is sticking with it's Frame Mode. And maybe this is why we're not seeing an AVC-HD DVX (using standard memory cards and a long GOP). What do you guys think?

[Barry_Green]

Who's the source of this information? It's just hearsay until we get an actual source. Is it from an official Sony representative? Or is it from Steve Mullen?

If this is true, and from an official Sony rep, then the V1 is simply not fit for purpose. I doubt Sony would screw up that badly, but who knows. The US model doesn't appear to be any different. I would tend to doubt it until seeing official confirmation.

Besides, the whole concept seems to be bizarre. The Canon records 24F and 30F with no problem, so if there were any truth to that statement Canon would seem to have completely defied it. Which would mean it would be a limitation in Sony's encoder, and only in Sony's encoder. That would have no effect on any other manufacturer, any other encoder, or certainly not on AVC-HD which is an entirely different compression algorithm.

Furthermore, it really makes no sense -- why would it be able to handle 50 interlaced fields per second, but not 25 progressive frames? Progressive should be easier to encode than interlaced.

Sounds like complete hogwash, but who knows? Let's get some sources and some official statements.

And I'm not prepared to rule out user error either; I'd sure like to see someone like Adam Wilt try one of these and see what circumstances the "oil paint" effect comes into play, and under what circumstances it disappears.

[Chenopup]

I'm with Barry

I think it's going to require some pro testing to really determine whether there is an issue. I can see the "paint" look in stills but when output to both HD and SD monitors, the only thing I can see is the slight softness change from the interlaced to progressive image. Nothing that at this point would even concern me based on a visual perspective.

I don't have my DVX here at this point or I would test 60i - 24p images with that as well.

the progressive image IMO has always been softer. In the DVX, HVX, XDCAM, Z1 (simulated of course) and now the V1u. Truth be told, I haven't seen this type of debate over the image difference between 60i and 24p on any of the other cams and wonder if it's because no one's looked that deep before.

Overall, I'm very impressed with the V1 and think that if there is a problem (again which I don't see to be that big of an issue when taken out to monitor), then Sony will step up and fix it.

Best to wait and see

cheno

[philnerd]

Well I spent more time looking at Brett's screen caps and found that the progressive frame washes out details in some ares and actually sharpens details in others. In Brett's grabs the red shed, especially the front side with the single door, has detail rubbed out big time. But finely detailed and tiny branches in the right of the frame are actually crisper and better defined in progressive.

Its really odd. I'm glad I got the XH A1, that's for sure.

[Barry_Green]

Okay, I have confirmed via a trusted engineering genius with an American V1U that yes, the US V1U is also affected with the "oil paint" effect, and it happens in both 24p and 30p. The effect gets more pronounced under gain. Pictures, article, and definitive insight into the issue should be forthcoming in a few weeks.

[John Trent]

Don't know if anyone cares anymore, now that Barry has confirmed the problem, but here's the page where I read it.

http://www.sonyhdvinfo.com/showthread.php?t=8195&page=2

I still question HDVs' GOP and 24p, mainly because it's not been done, at least not true Progressive and not without problems. Adam Wilt says the JVC HD100 "sticks" for 6 frames at times and the Canons' drop resolution from 700 to 540 lines using it's voodoo frame mode. Also, I can't forget that Jan said long GOP doesn't do 24p any favors. The XDCAM isn't as bit-starved and has no issues I've heard of, maybe Sony will record to Blu-Ray discs with a Z2 model, up the bitrate, and avoid the problem. I wonder if the CMOS sensors play any part in this. The Clear-Vid's are noisier than ccd's and, I've read, have issues with diagonals.

[Barry_Green]

But the Sony isn't recording 24p, it's recording a 60i stream using 2:3 pulldown. So nothing changes about the GOP at all. It should actually be a tad easier to record the 24p-within-60i than it is to record raw 60i; at least there's some field redundancy in the 24p due to the 2:3 pulldown.

Besides, it has *nothing* to do with the encoder -- you can see the effect on the live uncompressed outputs. It's something happening in the DSP, not the HDV recording. I mean, of course the effect is visible in the HDV recording, but it's not the process of recording HDV that's causing the problem, the problem exists before the footage ever hits the encoder.

The only advice I can say right now is: avoid gain like the plague, as gain appears to really accentuate the "oil paint" effect. And if you want to avoid the "oil paint" effect entirely, shoot interlaced.

[TimurCivan]

i take its that mushyness that appears on any "similar" tone or color?

[Barry_Green]

The cause and effect are being explored and will be explained in detail when they're known and explored.

[TimurCivan]

hmmm. too bad, seemed like a nice camera.

[Barry_Green]

It may still be. The cause and effect are yet to be fully explored. It may turn out to be easily controllable, or that the effect only manifests itself in easily-avoided situations, etc.

[digiteye]

It may still be. The cause and effect are yet to be fully explored. It may turn out to be easily controllable, or that the effect only manifests itself in easily-avoided situations, etc.

Barry, based on your experience, does it make sense to expect Sony to recall cameras to replace a physical part (DSP circuit or something) in case if such a part would be identified and confirmed to be the reason for the problems? Similar thing happened thousand times with Apple batteries, various parts of certain car models, etc., so in other industries it is not unusual - and a very good approach to regain customers trust.

I am asking this because V1E is quite a usable camera and has its advantages vs other models. For me the size, the big LCD, HDMI out and easy operation are factors why I tend to keep it.
However if the PQ will remain an issue and Sony might only improve in future CMOS cameras I may consider going for the full refund that is offered now.

Thank you for sharing your knowledge about it.

[John Trent]

But the Sony isn't recording 24p, it's recording a 60i stream using 2:3 pulldown.

Besides, it has *nothing* to do with the encoder -- you can see the effect on the live uncompressed outputs. It's something happening in the DSP, not the HDV recording.

Barry, doesn't the HVX record 24p (at 1080) in the same way?

I wasn't aware that you can see the effect on the live uncompressed outputs. That changes everything. I'm glad to hear that.

Could you explain why progressive is easier for the encoder. I thought it would be harder because instead of two separate passes of 540 lines, as in interlacing, it's doing 1080 lines all at once. To use what is probably a bad analogy here -- I'd rather make two trips carrying two 50 lb. bags up a hill, then one trip carrying one 100 lb. bag. I guess it doesn't work that way, huh? Still, I'd like to learn why.

Thanks for your help.

[Barry_Green]

Barry, based on your experience, does it make sense to expect Sony to recall cameras to replace a physical part (DSP circuit or something) in case if such a part would be identified and confirmed to be the reason for the problems?
Voluntarily? No. Even when the PD150 had a demonstratable sound hiss, and Sony fixed it in later models, they made owners PAY $150 to have the sound fixed in their existing PD150s.

The fact that they started offering full refunds is really quite forward of them. But I would say this: if the thing doesn't do what you want, don't buy it and hope and pray for a fix in the future. They've already said they won't fix it. So even if they upgrade a future model, don't expect that upgrade to be made retroactive.

But they may surprise us too. I'm just saying -- make your purchase decision based on what's offered today. And the landscape today is, as I read it: it is what it is, and it's not gonna change, so if you don't like it don't buy it, and if you already bought it and don't like it, they'll give you a refund (Sony Europe will, apparently; I haven't heard of any similar statement from any other Sony territory).

Similar thing happened thousand times with Apple batteries, various parts of certain car models, etc.
But those are usually either safety issues, or proven defect warranty issues. Sony is saying that this is *not* a defect, just like JVC said about the split-screen issue. SSE clearly was a defect for many of us, but I believe JVC had to take the stance that it's "not a defect" and state that up front as a legal protection against being forced to take large numbers of returns. Sony appears to be saying that the oil paint thing is "not a defect" and it may be for similar legal-protection reasons. Which makes it all the more intriguing that they do offer refunds in Europe, as that would seem to me to establish a bit of a precedent. Then again, I'm not a lawyer and I don't even play one on TV, so this is just speculation.

I am asking this because V1E is quite a usable camera and has its advantages vs other models. For me the size, the big LCD, HDMI out and easy operation are factors why I tend to keep it.
The V1E has a lot of things going for it. And apparently it is unaffected in interlaced, so if you want a 50i camera and enjoy the features it offers, don't even worry about it. Just be aware that if you bought it for 25p usage, there are apparently issues (the full extent and nature of which are hopefully going to be documented soon) and those aren't going to be "fixed" apparently.

However if the PQ will remain an issue and Sony might only improve in future CMOS cameras I may consider going for the full refund that is offered now.
All I can do is guess, and I would guess that the existing model that you have wouldn't be updated. They tried that, they offered a "fix", it introduced new problems, and now apparently allegedly insert-appropriate-disclaimer-here they're saying "no more fixes, it is what it is." If a fix were to come out in the future I wouldn't expect it to be retroactive, but they could always surprise us. JVC had a few defects in their HD100 like the "mode changing" issue and the battery readout incompatibility, and they offered a "free upgrade" to the HD100A to repair that. Gotta give 'em props for taking care of existing customers, for free. That can't have been cheap, and they offered it for every one of their existing units no matter when it was bought. Sony may decide to do that too. Or they may not. Or they may take the tactic of what they did with the PD150. Time will tell.

In any case, if you like the V1E and you don't use 25p, why worry about it? But if you bought it specifically for 25p, I'd say that planning on keeping it and hoping that maybe they'll fix it for you in the future for free is a lot like a woman marrying a man who isn't what she wants, but she secretly hopes that she can "change him." Then again, you may find that you don't encounter the oil paint effect, or that you don't mind it, or that you can't even tell the difference. If any of those apply, then again why worry about it (unless maybe it becomes an issue for a future ebay sale or something when you decide to upgrade?)

[Barry_Green]

Barry, doesn't the HVX record 24p (at 1080) in the same way?
Yes. But the point wasn't to say "this is better than that", the point was to say: the V1U isn't recording 24p any differently than it's recording 60i. There's no new recording process going on. Whether you shoot 24p or 60i, it's still going to record it as a 60i stream, so how could it possibly be "overloading the codec" or anything? It's not a new recording. On the Canon it's different, the 24F is actually recorded differently than the 60i, so if there were a case to be made for that I'd say the Canon would be the more likely candidate (but the Canon does fine with its 24F).

All I was getting at was: the way it records its 24P is no different than the way it records its 60i. So it's not by virtue of it being sourced at 24p that's causing the problem.

To use what is probably a bad analogy here -- I'd rather make two trips carrying two 50 lb. bags up a hill, then one trip carrying one 100 lb. bag.
But would you rather make two trips carrying two 5lb bags up that hill, vs. one trip carring a 10lb bag? I'd way rather get it over with in one trip if I can! :thumbsup:

[Simes]

Canons' drop resolution from 700 to 540 lines using it's voodoo frame mode.

The A1 goes from 660 down to 590 in 25F according to Tom Roper using the MTF50 test. Isn't that about the same as the HVX Barry?

I'm purely talking technical resolving resolution - not perceived image or anything. So I would say 'progressive' on HDV is possible (Based on the little I've done so far with the A1)

[Juan Diaz]

For the sake of clarification, here's the link to the MTF50 test referred to by Simes:

http://www.dvinfo.net/conf/showthread.php?t=80254&highlight=MTF50

[Barry_Green]

Yes, that's about the same, and the side-by-side pics bear it out.

[Jason Rodriguez]

I'm looking at those pics, and while I can't say for certain what the issue is, it seems to me that the progressive images (at least in the case of the workshop images that were posted) have a lot more fine detail that is being mushed by the MPEG encoder, where-as the interlaced images are naturally softer, so there's less detail to deal with, and the MPEG encoder is not exhibiting mosquito noise, etc.

Of course did someone say that even the HDMI output are exhibiting the same issues? If so, then I guess this isn't the same thing that I'm thinking about.

But to me this "paint" effect, at least on the edges, looks like noise reduction on mosquito noise that an MPEG encoder (or any DCT algorithm) would exhibit on hard-to-encode areas, not sensor-noise from the CMOS itself. Also there could be some interpolation issues in creating a normal image from the diagonal pixel pattern of the sensors that is creating false detail patterns that are creating too much pixel-to-pixel variations which again would "break" the MPEG encoder, or at least cause excessive mosquito noise issues (and the DSP might be trying to run noise reduction on that stuff).

I'm not sure if I would call that a "defect" per-se, just might be a limitation of the way they implemented the technology in the camera. In this case their diagonal interpolation scheme might be creating fine pixel-to-pixel variations in edges and other high-detail areas that is creating to much "detail" for the MPEG encoder to handle, or is creating false noise patterns that is causing the DSP to run noise reduction on those areas when it doesn't need to.

[Jason Rodriguez]

Actually, to see if the above is true, I would try to defocus an image and see if you still see the same artifacting, or if the artifacting is only in areas of high frequency detail or edges.

In other words, shoot a scene out of focus, and slowly roll it into focus . . . check and see at what point the "paint" affect starts and on what type of detail it starts showing first. Maybe even shoot a scene that would have one item in focus at one point and another item out of focus (so rack focus), so you can see if the "paint" effect is removed from an object as it gets softer.

Just some ideas to see if the above theory about too much false high frequency detail from either interpolation or the MPEG encoder getting overloaded is true.

[xray]

Canons' drop resolution from 700 to 540 lines using it's voodoo frame mode.

Tom Roper spend some more hours doing tests, and these are the results:

Updated Tests: (01/01/2007) XHA1, 60i and 30F

--------------------------------------------------------------------------------

XH-A1 60i: (01/01/2007)
824.3 Lines Horizontal (MTF50); 810 Lines Horizontal = Limit@Nyquist
674.8 Lines Vertical (MTF50)

XH-A1 30F: (01/01/2007)
823.6 Lines Horizontal (MTF50); 810 Lines Horizontal = Limit@Nyquist
592.2 Lines Vertical (MTF50)

Previous Reported Result for XH-A1 24F:
810 Lines Horizontal (Limit@Nyquist)
590.4 Lines Vertical (MTF50)

[Simon Wyndham]

It did surprise me actually that the US V1 was said not to suffer from the issue. When I originally discovered the problem a lot of people couldn't see the problem, Steve Mullen included, until it was really pointed out as being rather more than just being 'a bit softer'. I was assured that the issue had been taken care of. I can tell you that Sony did take the issue very seriously. But perhaps the problem is too far entrenched in the hardware that it is unfixable. This would be a great shame as it is a fantastic little camera in all other respects.