Panasonic Removable Lens Four Thirds HD AVCHD Camera

You guys are serious? You're considering this the Scarlet Killer? Isn't that what they said about the D90, and then the 5D Mark II?

Now in all consideration, the GH1 and the specs it gives are amazing and make it worthy of a possible Scarlet Competitor, but I'm just shocked its being called a Scarlet Killer
 
Zacatac said:
You guys are serious? You're considering this the Scarlet Killer? Isn't that what they said about the D90, and then the 5D Mark II?

Now in all consideration, the GH1 and the specs it gives are amazing and make it worthy of a possible Scarlet Competitor, but I'm just shocked its being called a Scarlet Killer
Click to expand...

It's very easy to be a killer of something that simply doesn't exsist... I'm more than tired to hear about cameras that are only a project that was announced 3 or for years before

Where is Scarlet now??????
There are only: a mock-up, a 3d model, a list of specifications, a forum and, obviously, a new delay announcement

Very very very easy to be a Scarlet killer

Gary
 
Filmmaker's Gang has be confused with someone else. That graph does say it all. It definitely says why an HD image downconverted to SD will virtually always be better than an SD image from an SD camera. And by extension why a GH-1 likely exceed the quality of any 2k sensor camera. Now a 4k sensor camera is a different story. The Scarlet comes in 3k, 4k, 5k, and 6k configurations.

Other than the graph posted, and the MTF of the recoding mechanism...I'd worry about lighting, filters, and audio. And of course the script!

1236960938.jpg
 
Last edited:
DavidNJ said:
Filmmaker's Gang has be confused with someone else. That graph does say it all. It definitely says why an HD image downconverted to SD will virtually always be better than an SD image from an SD camera. And by extension why a GH-1 likely exceed the quality of any 2k sensor camera. Now a 4k sensor camera is a different story. The Scarlet comes in 3k, 4k, 5k, and 6k configurations.

Other than the graph posted, and the MTF of the recoding mechanism...I'd worry about lighting, filters, and audio. And of course the script!
Click to expand...
man.. agreed on your finds not w/ your conclusions though.. well.. not all :D
 
filmmaker's gang said:
man.. agreed on your finds not w/ your conclusions though.. well.. not all :D
Click to expand...

All you have to do is take some EX3 footage (vimeo has some), HPX300 (this site), 5DM2 (vimeo), SX1 (vimeo), and Henry Olonga's HR11 through HDMI to Cineform (vimeo). You can add 35mm adapters if you'd like.

Examine them in a binary method, side by side. a to b, b to c, a to c, etc. It become obvious fairly quickly. Then add the presentations from Panavision's John Galt and Canon's Larry Thorpe, and maybe the Panasonic presentation on MPEG2 vs H.264.

I am always responsive to new information. However, if the GH-1 a) processes a 4k x 2.25k image, b) outputs the never compressed image out its HDMI port I will call this a defining camera.

However, there is room for full frame sensors, full sets of video controls, better in camera audio handling, and in-camera recording of high bandwidth 10-bit 4:2:2 H.264 to compact flash or even a hard disk array or solid state storage. In the next two years todays cameras will seem as obsolete as most DV cameras did in 2004 after the DVX100 established itself.
 
Gary Senda said:
Only the newest lenses lack the aperture ring... you can buy tons of used lenses with the aperture ring. They are a lot cheaper and, if in good conditions, they are much better than the actual lenses for digital cameras

Gary
Click to expand...

Thanks for that, just wondering do you know of any old lenses that would be wide with the 2.0 crop factor? As most of the new ultra wide zooms I see have no aperture ring so if you want really wide might only be able to go with newer micro or four thirds lenses and an adapter? Also does anyone know whether there will be an option for 1/48th shutter as I have seen mentioned that 1/60th is the lowest, yet if this works similar to an hv20 (60i/24p) cmos it should be possible? Cheers
 
I wouldn't call the GH1 a scarlet killer even if it did kill the chances of me getting a scarlet. :) It just doesn't seem necessary to own the scarlet now for a lot of us. We just wanted better than hdv recording, better than 1/3 inch DOF, solid state cheap recording, interchangeable lenses. So now that we have that for under $2000 scarlet just doesn't seem that far away from the competition. If this same camera was released in a prosumer body with it would have cost $5000 easy.
 
Right. You definitely can't compare a 1.500$ consumer camera to a much more professional camera that will cost three, four of five times more.

But if the GH-1 ships with a HDMI out, things will look different.

Also Panasonic now owns all the ingredients to produce a professional scarlet on their own: AVCIntra & the GH-1 sensor. They could easily take a HPX300 body and stack a GH-1 sensor on it. THAT would be a real scarlet killer for about 7 or 8K. (with a f2.8 zoom lens included) If there is a market for such a camera they could present it pretty fast.
 
GH-1 vs. Red
Subtitle: Red fans needn't worry

The red stores a 288 Mb/s data stream similar to Cineform. The data is a raw format that demosaics the Bayer filter in processing. The GH-1 could store an uncompressed image, but you are starting with one downsized to 1080p or 720p and demosiaced. You wouldn't have the same processing capability.

Sensor is actually larger on the GH-1 compared to the Red One, but the Red One has about 10% higher resolution. If you look at the second image that Isaac posted, and wonder why it looks that way, look at the last two graphs I posted.

The glass used for that picture isn't sold yet on their site, but is probably in the $3-5k range. A typical set of cine primes would probably cost about $20k or so. They rent for $700-800/day.

How to get close:

1) GH-1 on a really good set of legs

2) Good 50mm f1.4 35mm prime...say $2000. The Leica 50mm F1.4 is $3600.

3) Capture to Cineform (10-bit, 4:2:2) or with the Convergent Devices nanoFlex XDR and 'maybe' the Panasonic HGR20 AVC Intra. In the first two you will have about 160Mb/s.

4) Light your subject for no gain and F5.6 or so.

5) Set up the camera's color setup, white balance and focus carefully. I find a target helps.

6) I would also throw on a light (1/4, 1/2) Black Diffusion filter...but that's me.

My guess is 90% of the people here would then have a hard time picking the image from that Red shot. You will have around $10k in equipment invested.
 
The scarlet is a great cam it seems, but yeah, it's definitely not geared toward guerilla style filmmakers. Due to the simple fact, not only is the camera pretty heavy on the wallet, but your system is gonna need a major upgrade to edit. The gh1 brings me closer to home.
 
couldn't agree more. One problem I have with red is that it is still a very local player. Service should be better with panasonic, especially in a third world coutry. So I would prefer to get 90% of image quality but better service. Forgive me if I'm wrong.
 
GH1 is not a videocamera, it is only the first real attempt of hybrid camera. 5D is a photocamera with some limited video capabilities. GH1 will not substitute my 170, but it can be its ideal integration in several situations.
For some projects I will use my 170, for other projects I think the Gh1 will be much better. For other I could use both.

In addition it is small and light, the perfect camera to carry with you without choosing between video and photoes and without limiting your movements. It is a very flexible camera, easv to carry with you and this means you can be more productive

Gary
 
Gordon Prince said:
This isn't true. Quite the opposite.
Click to expand...

You are correct...I was getting the small Scarlet and the Red One confused...

The point was...the Red One was an example of the GH-1 with HDMI taken to extremes...better sensor, higher oversampling, better glass, much better recording.
 
David, these things happen. The point is not to let an untruth or even half ones slide by.

Here's my typo -- not exactly false but anyway :-)

Emanuel said:
I think David is referring Scarlet (2.3") not the RED ONE (S35). Just a sort of typo.
Click to expand...

It is 2/3". We shouldn't let it as 2.3" or can bring confusion towards 1/2.3", the SX1 sensor size. Not the same, closer to 1/3" DOF than 2/3" actually.

Apart the full frame sensor size discussion, on your comparison with the GH-1 plus HDMI, I must admit that the combo can bring a different value-added but the interlaced encoding remains or am I wrong?
 
[FONT=&quot]The image is in a 2:2:3 pulldown. That is two frames are pulled down to on progressive frame and 3 progressive frames are pulled down to 2 24p progressive frames. Software pulls that out.

I will check with Cineform and Convergent Designs how they handle it. I believe both do. If I'm correct, the Convergent Designs devices would save the output as native 1080p24 in MPEG-2 I-frame only at 160Mb/s. That is only 5:1 compression. By comparison 17Mb/s, with the 3:2 pulldown removing 20% of the bandwidth is 59:1 compression. The use of a long GOP in contrast to I-frames may create a defacto 50-100% increase in effective bandwidth and reduction in compression, specially for relatively low movement scenes.

For comparison, DV is around 5:1 compression. The Canon 5DM2 is 13:1, and has the same GOP advantage. Giving Long GOP a 2:1 advantage (generous to long GOP) that would give Convergent 5:1, DV 5:1, AVC Intra 100 6.6:1, 5DM2 13:1, and GH-1 22 or 31:1.

H.264 is better than MPEG-2, but it varies by bandwidth. Lets say it is a log scale equal to 2x at 13Mb/s and zero at uncompressed, say 1.2Gb/s to make this easier.
Now, in our napkin analysis virtual MPEG-2, AVC Intra 100 is 6.6:1, the 5DM2 8, and the GH-1 12 or 15. On this scale XDCAM would be 13.5..

Confusing?

Let me simplfy, compressing [/FONT]

  • [FONT=&quot]DVCPro50: 3.5:1 (in case you were wondering why SDX-900 footage looks so fantastic)[/FONT]
  • [FONT=&quot]Red One: 4.2:1[/FONT]
  • [FONT=&quot]Flash XDR 5:1 I-frame or Long GOP[/FONT]
  • [FONT=&quot]AVC Intra 100: 6.6:1[/FONT]
  • [FONT=&quot]DV: 5:1[/FONT]
  • [FONT=&quot]5DM2: 8:1[/FONT]
  • [FONT=&quot]DVProHD (720p24@40Mb/s): 9:1[/FONT]
  • [FONT=&quot]XDCAM EX: 13.5:1[/FONT]
  • [FONT=&quot]HDV 1440x1080p30: 13.5:1[/FONT]
  • [FONT=&quot]GH-1 720p: 12:1[/FONT]
  • [FONT=&quot]GH-1 1080p: 16:1[/FONT]
  • [FONT=&quot]HDV 720p60: 20:1[/FONT]
[FONT=&quot]This napkin analysis probably overvalued long GOP, however, you can see the 5DM2 isn't shabby. Cineform would be the same compression as the Flash XDR but 10-bit and more lossless compression.

I have made comments demoaning that AVC Intra is I-frame only. Had it been long GOP on this napkin analysis it would have had a 6:1 compression.

Anomaly? The HMC150 recording 720p24 to AVCHD without pulldown would be 4.7:1, much better than the HPX170. My observations seem to find the frames similar, maybe showing the limits of the MTF in the lens and undersampling sensor. Or it may just be a weakness of this analysis overvaluing long GOP and AVCHD.

But the key is oversampling...as shown by my last curves. As the Canon shows, once you get in the MPEG equivalent 10:1 range the MTF curve of the compression lets texture come through. The Henry Olonga's HR11 via HDMI to Cineform clips show how important the MTF of the compression is.[/FONT]
 
Last edited:
hey man.. here you go again.. nice.. :D did you take a look on that though?

scroll down to the page's end.. comments.. mr wilt:

"Certainly the reverse telecine removes the interlaced lines, but interlaced 4:2:0 color encoding is fundamentally broken. The hummingbird image was reverse-TKed; the luma is all OK, but the chroma has that “sawtooth” edging on it still."

“Isn’t there any solution for removing that ‘sawtooth’ edging?”
Aggressive chroma blurring is the only thing I’ve been able to do. If you’re using FCP or FCE, try the “Channel Blur [AJW]” filter set to YCrCb (YUV) mode, and fiddle with Cr and Cb blur levels. Download “AJW’s Filters. zip” from http://www.adamwilt.com/downloads/
Actually, a vertical-only (or predominantly vertical) blur might be more useful. If you feel like playing, edit the filter and fiddle with the aspect ratio parameter in the BlurChannel calls."
 
You must log in or register to reply here.
Back
Top