HPX500 vs. HDX900

[skillimon]

Trying to sort out a new camera purchase.
Setting aside all the issues related to P2 vs. tape, how do these two cameras stack up,camera to camera? Picture quality, low-light capability, easy of use. Any experiences or comments helpful.

[hemophilia]

bump.

Anyone with info on this? I think I've done a fairly thorough job searching the board for such info; and the Panasonic site is TRAGICALLY uninformative. They make very little effort to differentiate their product lines and explain the differences. A bunch of comparisons were made with the 2000, but not with the 500 vs. 900.... so what are the differences, besides tape?

-Kevin

[marco0782]

As far as I know, the HDX900 does not do variable framerates but it does have an HD sensor whereas the HPX500 uses an SD sensor like the HVX and achieves HD resolution through pixel shifting.


Marco

[hemophilia]

Yeah sort of in answer to my own question (and skilimon's) I've dug around a bit more and found some answers:

http://forums.creativecow.net/thread/193/866807
(Barry Green's discussion on Creative Cow)

http://ear.vo.llnwd.net/o2/3000/3000-1961.jpg
(A basic specs comparison chart of most of the Panasonic line.)

http://www.simonwyndham.co.uk/panasonic-hpx-500-review.html
(A review that helps differentiate it from its more and less expensive brethren)

-K

[smelni]

As far as I know, the HDX900 does not do variable framerates but it does have an HD sensor whereas the HPX500 uses an SD sensor like the HVX and achieves HD resolution through pixel shifting.


Marco


The hvx is not an SD sensor - pixel shift doesnt make resolution from nothing - its a smart way of having less noise and more resolution.

[marco0782]

The hvx is not an SD sensor - pixel shift doesnt make resolution from nothing - its a smart way of having less noise and more resolution.

Yes it is an SD sensor. An HD sensor is slightly above 2 megapixels whereas the HVX sensor is around 0.6 megapixels.

Pixel shift is a smart way of achieving HD-like results with a sensor that has less pixels. I don't see that noise has anything to do with it because the HVX is slower and noisier than the Canon HV20 with its full 1920x1080 sensor.

Pixel shift works pretty well but it's no substitute for the real thing. The image from an HV20 is sharper, in my opinion, but my HVX holds up very well against the XLH1.


Marco

[mikkowilson]

The HVX200 and the HPX500 both have HD sensor blocks, no question about it.

Pixel shift simply combines 3 sensors into one higher resolution one.


- Mikko

[marco0782]

From DVinfo.net (http://dvinfo.net/conf/showthread.php?t=88095):

"It's worth noting that this resolution increase applies mainly to luminance, and is most effective for a subject with low colour saturation. For a highly saturated subject, pure red lit, say, all the output will be from the red sensors, none from the green or blue, so the pixel shift technology can't have any effect.

Since the human eye is more sensitive to luminance resolution than chrominance, and most TV systems have more limited chrominance resolution anyway, it should generally be seen as a good thing."

[mikkowilson]

And that varies from a single HD chip how exactly?

- Mikko

[smelni]

Yes it is an SD sensor. An HD sensor is slightly above 2 megapixels whereas the HVX sensor is around 0.6 megapixels.

Pixel shift is a smart way of achieving HD-like results with a sensor that has less pixels. I don't see that noise has anything to do with it because the HVX is slower and noisier than the Canon HV20 with its full 1920x1080 sensor.

Pixel shift works pretty well but it's no substitute for the real thing. The image from an HV20 is sharper, in my opinion, but my HVX holds up very well against the XLH1.


Marco


I could cite sources of why it is an HD chip - or I could post a resolution chart and show that it indeed shoots HD images - which would be impossible with an SD chip - but this question has been done and redone - just do a search here.

[marco0782]

"And that varies from a single HD chip how exactly?"

In that it only produces HD resolution for luminance. If you are seeing an all red image it will be in SD. With a 3-chip HD sensor there would be 1920x1080 red pixels.

[mikkowilson]

aah.. a 3 HD-sensor design, sure, but in a camera this size, the pixels on the sensors would be tiny, and dynamic range would suck. That's why the only cameras with 3 fully HD chips are 2/3" cameras, where the size increase allows for bigger pixels.


I was asking how pixel shifted HD CCD blocks differ from a single HD chips though - as in, what make the HV20 with 1 CMOS chip any higher rsolution than the HVX with a pixel shifted CCD block for HD?

- Mikko

[Barry_Green]

Yes it is an SD sensor. An HD sensor is slightly above 2 megapixels whereas the HVX sensor is around 0.6 megapixels.
There's nothing about this statement that's correct though. It shows a fundamental misunderstanding of how sensors are read and interpreted. You do not ever read one chip -- individual pixel counts on a single chip are largely irrelevant. It's the total combination of all the chips being read together that produces the final image. And in that context, the HVX resolves a 2K image; the imaging block is a 2K imaging block (regardless of how many pixels are on each individual chip).


I don't see that noise has anything to do with it because the HVX is slower and noisier than the Canon HV20 with its full 1920x1080 sensor.
Marco, come on. This statement is so incredibly easy to disprove, why do you even say things like this? The HV20 is an ISO of 80. The HVX is an ISO of 320. The HVX is two full stops faster than the HV20. To make an HV20 "faster" than an HVX you'd have to pile in 18dB of gain, and at that point it's far noisier than the HVX.

Any perception of the HV20 being "faster" is probably driven by the fact that it uses such a prodigious amount of electronic gain, up to at least 27dB...

[smelni]

Barry to the rescue :)

[DerrickTempleton]

I was getting really confused there...thanks Barry

[marco0782]

The HV20 is an ISO of 80. The HVX is an ISO of 320.My mistake, it is the XH-A1 that is faster and less noisy even though it has more pixels on its sensors.


It's the total combination of all the chips being read together that produces the final image.I'm eager to know how the green and blue sensors contribute any data besides luminance when recording a red object.

[smelni]

what else is there to contribute from the green and blue except luminance no matter what the size of the 3 ccds - it wouldnt matter the resolution - the offset is still there for the luminance

[mikkowilson]

Well, ALL sersors only detect intensity (luminance). The Prism is simply a color filter to direct the light to the correct sensors.

Compare to how a single chip camera works, using a bayer filter:
http://www.i-cubeinc.com/images/3shotcm_files/3shotcm_figure1.jpg
The reported resolution of the chip itself is full HD, but each color gets way less resolution!
Those colors are of course then processed into the component video signal.
Those 3 color "layers" match the 3 CCDs of a 3CCD camera, of course there arn't wasted pixels for each color, but simply larger (more sensetive) pixels that are positioned (using "pixel shift") in the right spots to provide the colors, but still the combined overall luma grid.

This is why ALL the 1/3" 3CCD cameras use pixel shift, it just makes sence.



Of course when all this becomes a component video then it's all "brightness of each color" signals for RGB processing & compression .. so if you only get one color into the camera, then you'll only get one color thorugh the compression anyway, so what loose light sensitivity over it?

- Mikko

[marco0782]

Thanks for clarifying, that makes sense.

But if you are recording a deeply saturated red object, how do you get 1920x1080 resolution if two of three sensors are not receiving any light at all? You don't. And so in practical terms unless all three chips are being used to their fullest you are not getting true 1920x1080 resolution.

[mikkowilson]

True enough. But it would be pointless to capture more than can be recorded by the recording codec.

A 4:2:2 codec (Like DVCPRO-HD) will only record every other color sample anyway.
And a 4:1:1 or 4:2:0 codec (Like HDV) will only record every 4th color sample.

So even if you could capture a full 1920x1080 red signal, the codec wouldn't be able to store all those pixels anyway.

- Mikko

[marco0782]

With 4:2:2 there would be a loss of color resolution but not a loss of luminance resolution, correct? So therefore, if only one chip receives light the luminance resolution will be standard definition too.

This might be helpful here:


Color resolution is measured in "lines of resolution" over a given distance span just as is regular resolution (luminance resolution) where each "line of resolution" equals one color change.

Changes between a light and a dark shade of the same color are luminance changes and are not affected by limitations in the color resolution.

http://members.aol.com/ajaynejr/vidcolor.htm

[mikkowilson]

Aah yes, this I think must be the point of confusion.

Luminance is simply the luminance of each color - all luminance is of one or more colors.

Notice that there is no "Luminance" sensor, just a sensor for (the luminance of) each color.

RGB video is simply the lumiance of R, G and B - there's no seperate luminance channel.
Then when you convert it to Y, Cb, Cr, (Component video) for recording, then you have Y - which is simply the sum of all the colors (but still dependant on them) and then CB and Cr which are just calculated signals to show "color difference" - It's just a different way of storing the data, but it's stil the same data.


Also keep in mind that the lack of blue and green on a red subject, doesn't mean that the resolution isn't there, it just means that it's value is "0", and thefore black - it's dark, but it doesn't mean it's not there!

- Mikko

[Barry_Green]

But if you are recording a deeply saturated red object, how do you get 1920x1080 resolution if two of three sensors are not receiving any light at all? You don't. And so in practical terms unless all three chips are being used to their fullest you are not getting true 1920x1080 resolution.
If you want to get pedantic, then yes -- if you imaged an object that was exactly the exact same identical hue of red or green or blue as the Beyer filter separates it out, or the prism splits the light into, then yes you'd get the resolution of the one sensor (or of the pixels in the Beyer pattern).

What are the odds of shooting that exact wavelength of light? Pretty slim.

In practical terms, there's always some crossover. Always -- or at least 99.99999999999999999999999% of the time. And when that crossover happens, it'll be detected by the other chips, and voila - the full imaging block is used.

Video isn't recorded in RGB anyway, every pixel undergoes a YUV conversion, where each individual pixel gets 60% of its signal from green, around 29% from red, and around 11% from blue (or thereabouts).

As Mikko demonstrated, a "true" 1920x1080 single-chip design isn't "true 1920x1080" for any color channel. It's 960x540 for red and 960x540 for blue.

So that's all meaningless. Just look at the images. The XHA1 is about the same speed (not more sensitive) and about the same noise level (just different texture to the noise) as the HVX is. But it has to use interlaced chips to get there. You get more pixels (but not any noticeable sharpness increase, and a different noise texture). You can't do true variable frame rates with it, because it's coming from an interlaced chipset. The HVX allows true variable frame rates because it's a progressive chipset, but because it's progressive they had to employ larger pixels to get comparable sensitivity and noise performance.

(BTW, nothing wrong with being pedantic, I get that way all the time. But there's a difference between observing what's theoretically possible to happen, and acknowledging what actually happens. I'll acknowledge that in most circumstances long-GOP works just fine -- but there are certainly some circumstances where it utterly fails. And I'll acknowledge that in the extremely vast majority of cases pixel shift works great, but there is a rare exception of shooting under pure single-color conditions where it won't work).

[marco0782]

There could be crossover 99% of the time but if it happens in certain areas of the frame and or in varying degrees of exposure your resolution will still be somewhat compromised, right?

Why is it that when I film someone wearing a red shirt the shirt is pixelated? Is this 4:2:2's fault or pixel shift?

[Barry_Green]

There could be crossover 99% of the time but if it happens in certain areas of the frame and or in varying degrees of exposure your resolution will still be somewhat compromised, right?
Only on those elements where it's that single pure color.


Why is it that when I film someone wearing a red shirt the shirt is pixelated? Is this 4:2:2's fault or pixel shift?
Most likely 4:2:2. It's really noticeable on 4:1:1 or 4:2:0.

[Gabriel Bordas]

Hi,

The diference in image betwen hpx and hdx are:
dynamic range HPX 8.5 stops / HDX 10 stops
less noise in HPX
special gamma curve four night shooting in HPX, but HDX have a much more natural tones
HPX more sensitivity 1stop

HDX have the varicam CCD and less quality optical block
More control, beter DSP, less artifacting from pixelshift to go 1080 from a 720 ccd compared of HPX senzor. The crhomatic aberacion in HDX it four some low cost lens. not work in the big glases no need, ans so on.

I attach 2 test frames, first one a comparacion "out of box" and my one cinelike 2 setup, at the end some CC. Second one with difrent cinelike2 setup. no CC.
Canon HDCg series lens. The first image have a big mjpeg compresion.

Gabriel Bordas

[hemophilia]

Thanks Gabriel--

Finally someone getting the discussion back on topic. :)

A couple questions for you:

Which camera are those pictures from?

And where are you getting those stats (like the dynamic range comparison)? Is that something that someone else published somewhere? Or is that from your own experience with the two cameras?

It looks like I'll be using the 900 for a project next week (that I was originally asking about). But in general I'm very interested in the 500, b/c of P2, b/c of cost, and b/c of low-light, which I tend to do a lot of (night-time cityscape stuff).

Thanks,
Kevin

[hunter richards]

I am cutting a trailer for a feature that was shot with a Varicam with ziess glass and I just picked up an hpx500 with a fuji 17x lens today! After staring at varicam footage for the past 2 weeks and looking at a few quick rolls from the hpx500 today, I am blown away with the quality of the HPX500! I can tell you that the Varicam does have more dynamic range (in the highlights especially). The highlights in the varicam (and surely the HDX900) seem to roll off to white nicely. With that said, The HPX500 still has great useable dynamic range and is on par in terms of sharpness (the hpx500 might be sharper).

The funny thing with us film/video/data/resolution/pixel freaks is that we forget why and who we are actually making content for. I have made up my mind, that it does not matter if you shoot with an hpx500,f950,varicam,hdx900, ect. Your client or audience (narratively speaking) will not notice any difference. (for 95% of your productions, it might be different if your work is on a 100 foot screen) But what they will notice is "my production crew shot with a big HD camera" You guys know how silly it is.

The problem is in the producers and cast/crew/investors (even me). is that We make choices based on politics and marketing. The Hpx500 is not marketed as a flagship camera for panasonic. Thats the Varicam's job (and soon to be the HPX3000's) But anyone with their head screwed on can see that the difference in between the varicam (40k for the body alone) and the HPX500 (20k for a kit with 4cards, evf and lens), is not a 20-40k difference in aquisition quality.

You pay a lot for that V.

Well, This turned into a rant and I am sorry.

Bottom line:

Shoot with what ever you can, focus on lighting and composition more than datarates and pixels.

[mbrand]

Trying to sort out a new camera purchase.
Setting aside all the issues related to P2 vs. tape, how do these two cameras stack up,camera to camera? Picture quality, low-light capability, easy of use. Any experiences or comments helpful.

I use both cameras and own an HPX. There is no question that the HPX is a little noisier and softer then an HDX. You get what you pay for. The HDX is also better in low light, it's rated F11@2000 lux and the HPX is rated f10 but I think the HDX is better then F11. That being said I love my HPX because of its ability to shoot DV,DVCPRO, DVCPRO50 and DVCPROHD and variable frame rates, wow. You CAN overcrank (slow mo) an HDX by shooting 60P or 30P and applying software that panny provides called 'variable frame rate converter'. It looks great but you can only do 24P, 30P, and 60P, no undercranking and its time consuming. With an HPX you can playback in camera and see the effect instantly. The HPX sure is a bang for your buck with it recording versitality, 2/3" chip and mount, variable frame rates, pre-record function (HDX also has) and interval recording (for timelapse) which I love. I am happy with the HPX and look forward to upgrading to a 2000 or 3000 some day. Don't think I will go back to tape.

Mike Brand
mbrand@lafayettehillstudios.com

[walnutcrunch]

It's the little things that make a difference for me. I haven't been on the 900, but I have been on many many tape based cameras over the last 10 years and the biggest unexpected thing is the simple joy of hitting the little red button and instantly recording with no roll-up. That is just so insanely awesome especially in the type of work I do.

[hemophilia]

Mbrand--

HDX900 lower noise/more sensitive? Did you get that the right way? That's backwards from what others are saying (and intuitively backwards given the higher pixel density on the 900).

I really wish Panasonic would make a better effort to explain their product lines, making conversations like this pointless. They have a product range that spans $45K, and a "compare" feature on their site that is entirely worthless.

-Kevin

[TimurCivan]

I thin the HPX500 was the cleaner more sensitive one... i think?

[Brad Neal]

I thin the HPX500 was the cleaner more sensitive one... i think?

Since I just purchased 2-500's, I'm going to side with TimurCivan!! :-)

-Brad

[mikkowilson]

I yelled at both cameras for about 10 mintues after a shoot last week and the -500 burst into tears... it's definatly more sensitive.


- Mikko

[Barry_Green]

I believe the 500 and 900 are about the same sensitivity, but the 900 is cleaner in noise performance.