new Pannie HPX250 info

[David Heath]

Originally Posted by Sanjin Svajger

Folks that have lots of batteries, P2 cards and other accessories for the HPX170 are probably on the fence for the 250....

Fair point - I was basing my comparisons on buying a complete package from scratch.

Originally Posted by Glen Vandermolen

I don't know if the 250 has been approved for full acquisition by the BBC HD or Discovery HD, but since the HPX370 already is, and the 250 uses the exact same chipset and codec, I would expect the 250 to get approval.

You have to take the BBC list with a bit of caution - it used to signify "suitable for everything" now it's "passes the bare minimum". Hence the 370 (amongst others) may be allowed for some work, but frowned upon for other purposes. That said, I suspect they may be far less interested in the 250 now than they might have been a year or so back. They've bought a load of XF305s, and I don't see why they'd want to confuse matters by going onto another type of camera as well unless it offered a significant step up - and in the case of the 250 most aspects are comparable, but the servo lens is a significant step DOWN from the XF305.

They've also recently bought a large number of PMW500s, which together seem to show a preference for the 50Mbs XDCAM422 format.

When I looked at the video of the 250 at 12db, I thought it looked as clean as comparable XF305 footage.

I'd be a little careful for reasons mentioned earlier in the thread. The best guess is that it uses the same or similar chipset as the 370, and the "noise ghost" issues are well documented. Do you know what the noise reduction profile was set to in camera? The concern would be that noise is being minimised to look better at simple viewing - but give rise to other issues (the ghosts) that grab you further down the post chain. That said, the specs seem to imply it should be comparable to the XF305 for low light performance - but both to be trumped by the EX.

But for weddings I agree that the 130 is likely to be a more sensible choice than the 250 - in many respects they are the same or similar, but the 130 is cheaper to buy, and memory costs become far less of an issue.

[Glen Vandermolen]

I thought the 370 "noise ghost" was fixed with a recent upgrade?

And I'm not suggesting the BBC will go out and purchase HPX 250/370 cameras, but will allow independent producers to submit projects shot with these cameras.
And agreed, the XF300 lens is incredible.

[David Heath]

Originally Posted by Glen Vandermolen

I thought the 370 "noise ghost" was fixed with a recent upgrade?

I believe the "fix" was to offer an option of using less software noise reduction - effectively going back to something like the HPX300. The knock on is obviously that a "fixed" 370 is then noisier than a "ghostly" 370!

Since the 370 was introduced to counter complaints about higher than desired noise levels and the low light performance of the 300, I'll let you draw your own conclusions!! :-)

[Glen Vandermolen]

Originally Posted by David Heath

I believe the "fix" was to offer an option of using less software noise reduction - effectively going back to something like the HPX300. The knock on is obviously that a "fixed" 370 is then noisier than a "ghostly" 370!

Since the 370 was introduced to counter complaints about higher than desired noise levels and the low light performance of the 300, I'll let you draw your own conclusions!! :-)

Yeah, the P.A.P. filters, Type 1 and Type 2 and the "micro-jello." I was hoping Panny had ofered a more recent upgrade, but apparently not.

[Konstantin Kovalev]

50mbps long-gop is equal to 100mbps I-frame because the data is spread over several frames instead of a flat amount for each. In some circumstances (definitely not weddings, but who knows?) this plays an important role in compression because if there is too much change between frames (such as often happens in nature or if flash photography is involved) the first frame may get overloaded with information, and will leave about 10% of the data rate for the rest of the frames in the group.

They've also recently bought a large number of PMW500s, which together seem to show a preference for the 50Mbs XDCAM422 format.

Cash city my friend, how I'd like to live there, but $25k is too much for anyone but a network to consider.
...And that's only for the body.

Don't knock your heads too much about the issues surrounding the 370, it works great for what it is, I mean it's not like you're trying to capture a blockbuster on the darned thing, otherwise you're looking in the wrong section of the forum. With -3db and good light I can barely see some noise in the shadows, and on a typical HDTV with typical network compression you can't see anything until much higher gain levels are used. Don't be too critical of footage on a 27" 2560x1440 monitor from a foot away, it'll look awful.

[David Heath]

Originally Posted by Konstantin Kovalev

.......if there is too much change between frames (such as often happens in nature or if flash photography is involved) the first frame may get overloaded with information, and will leave about 10% of the data rate for the rest of the frames in the group.

This shouldn't be the case, and isn't with any good long-GOP codec.

The allocation of bitrate to ref frames and difference frames is dynamic, so the situation you describe shouldn't happen. On a static scene the diff frames don't get allocated much data (they don't need it), but with a lot of movement the ref frames get allocated less, the difference frames more.

Imagine two extreme cases, one with NO movement, the other basically a succession of random images frame by frame. If 25fps, 50 Mbs, and a GOP of 12, you could allocate the bitrate in two (extreme) ways

1. Allocate 25Mbs to each I frame, (two every second), and none to the difference frames.
2. Allocate 2Mbs to EVERY frame - difference frames getting the same as I frames.

1 would (theoretically) be optimum for a completely static scene, 2 would be optimum for the succession of unrelated images. In practice, real life situations fall between these two stools, but the underlying principle holds good. Less movement =less bits to diff frames. More movement =more bits to diff frames (and less to I frames)

It follows that for the same bitrate, and all else equal, a long-GOP system MUST be better in terms of quality than an I-frame only one. The debate starts when bitrates aren't equal - at what ratio can the qualities be said to be "equivalent"? The generally held figure is about 2-3x - that's why the equivalence between XDCAM422 and AVC-Intra 100 shouldn't come as any surprise - in spite of the 2:1 bitrate difference.

[Mark Donnell]

To me, it is highly questionable as to whether or not the HPX250 is an improvement over the HPX170. Panasonic could have chosen to dominate the ENG camcorder field by upgrading the three-CCD imaging block and the global shutter of the 170 to a true 1080p array, but chose instead to substitute its already-developed MOS-rolling shutter imaging block for the 250. Very disappointing, although good for those on a limited budget who aren't doing fast action video. Neither the new Canon C300 nor the Scarlet have incorporated CCDs or global shutters. Panasonic has missed a great opportunity here. Maybe they should consider an HPX350 at up to an $ 8500 price point with the aformentioned quality imaging block. Until then, I'll stay with my HPX170 until side-by-side tests demonstrate a rolling shutter without significant problems in fast-action shots.

[David Heath]

Originally Posted by Mark Donnell

To me, it is highly questionable as to whether or not the HPX250 is an improvement over the HPX170. Panasonic could have chosen to dominate the ENG camcorder field by upgrading the three-CCD imaging block and the global shutter of the 170 to a true 1080p array, but chose instead ...............

It's a nice wish, but think about it. What you want involves shrinking the size of the 170s pixels by a factor of well over 4 (well over, because of the guard spacing), if they'd stayed with CCD, what do you think the noise, sensitivity etc would be like? (The 170 is not exactly the most noise free camera in the first place.)

Basically, the choice was go to CMOS, develop bigger than 1/3" chips, or stay with 960x540.

As more and more displays become "full HD" (1920x1080) the differences between the 170 and cameras like the EX were becoming more and more noticeable - have you seen the two intercut? Hence the last option (stay at 960x540) gets ruled out. You'd need 2/3" to keep the pixel size the same and go to 1920x1080 - which becomes hardly practicable in this size and price of camera.

Hence CMOS.

[Mark Donnell]

So David, if I understand correctly, the increased sensor size necessary for a full-frame 1080 3-CCD camera would be too large for an HPX 170 sized or slightly larger camcorder, because the CCD sensor subunits are considerably larger than the CMOS subunits ? Or is it that with the rolling shutter, the CMOS chip doesn't need a full set of sensor subunits - instead, the same subunits are re-scanned slightly later with a different portion of the frame ? Sorry for the basic questions, but my knowledge of the detailed workings of an image block is rather limited.

[Mark Donnell]

Just one more slightly off-topic question, David. Several manufacturers are now making CMOS sensors with global shutters. Do you see this as the next major advance in the field of prosumer and professional video cameras ?

[David Heath]

Mark - the fundamental issue is that matters like sensitivity and highlight handling etc are heavily dependent on the size of individual photosites. Pack more onto a chip of the same size, and inevitably they become smaller - hence a compromise between resolution and sensitivity.

The only way to improve one without compromising the other is to move to a larger sensor or change the technology. The HVX171 is not the most sensitive of cameras as it stands - going to 1920x1080 and keeping 1/3" CCDs wasn't an option. Hence - bigger chips, or CMOS.

Bigger chips (you'd need 2/3" to keep the photosite size the same as the 171, whilst moving up to 1920x1080) mean bigger, heavier, more expensive lenses. Not really practical in a camera of the form factor/cost we're talking about here. Hence CMOS.

I'm not aware of any cameras currently that have CMOS/global shutter - do you have any references? Is it "the next major advance"? Probably depends what you do - some people get very disturbed by the effect, other people find it less of an issue. Even if not a global shutter, differing cameras have different readout times - mobile phone etc video shows it far worse than a true camcorder, even if both CMOS.

It will be interesting to see what the new Canon C300 is like in this respect. It uses a simplified readout compared to standard deBayering, which may make it relatively fast and hence may mean lessened rolling shutter effects. (Their readout method - see the C300 thread - is not new, what is new is that it's being applied to a sensor with the optimum pixel count for the technique, twice 1920 horizontally, twice 1080 vertically.)

[Sanjin Svajger]

Originally Posted by Mark Donnell

Several manufacturers are now making CMOS sensors with global shutters.

Who is making global shutter cmos Mark?

[Mark Donnell]

Two companies came up quickly on a Google search, and I believe that there are others as well. The first one, Teledyne DALSA is producing Falcon high-speed cameras with CMOS chips and global shutters (www.teledynedalsa.com), and the second one appears to be a chip distributor named Viimagic (CMOS image sensor, Imaging sensor). They are advertising a HDTV sensor chip that is CMOS with a global shutter, Apparently Kodak labs wrote some papers on CMOS with global shutters back in 2003, but the technology has been slow to come to market.

[Dom Stevenson]

The Panny looks like a great buy for those who've already taken the plunge with P2 cards. Not so good if you don't have them. Having said that, the card prices are not as bad as they were, and there seems to be a decent resale value in them.

It'll be interesting to see the camera side by side with the canon, but there is another excellent budget option. The mighty EX1 pops up on Ebay (often with very few hours on the clock) at half price with batteries, SxS card etc. The budget conscious can get a lot of camera for their cash, and even sell the camera on at little loss should they wish to move on to something else.

[Sanjin Svajger]

Originally Posted by Mark Donnell

Two companies came up quickly on a Google search, and I believe that there are others as well. The first one, Teledyne DALSA is producing Falcon high-speed cameras with CMOS chips and global shutters (Teledyne DALSA), and the second one appears to be a chip distributor named Viimagic (CMOS image sensor, Imaging sensor). They are advertising a HDTV sensor chip that is CMOS with a global shutter, Apparently Kodak labs wrote some papers on CMOS with global shutters back in 2003, but the technology has been slow to come to market.

Interesting... Looks like camera making companies will have to make their own sensors - they wont buy it from other companies...