Panasonic Removable Lens Four Thirds HD AVCHD Camera

[Isaac_Brody]

No more personal attacks please. If you can't play nice find another thread.

 

[Park Edwards]

i'm ready come june. if the gh1 isn't up my standards, just return it and go back to the d90. simple as pie.

 

[DavidNJ]

I can't imagine how the GH-1 couldn't be infinitely better than the Nikon D90.

What do you think of this clip:

 

[Park Edwards]

do you own or messed with one?

 

[DavidNJ]

SX1 or GH-1? Neither...although I have been chatting with people who have it and looking at lots of clips. When I started I was looking at the EX3, EX1, HPX170, HMC150. Looking harder, none are interesting any longer.

However, the SX1 probably sets the low end of what should be considered acceptable. '

I have another GH-1 question: is it limited to 4GB FAT32 files or does it have a mechansim for continuous recording?

 

[Kholi]

Good question, David. Event shooting still shouldn't be an issue if it's FAT32 Limited anyway because of the expense. However, it would still be nice to be able to record for extended amounts of time.

What other questions need to be answered?

I can agree with you on the HPX170-slash-prosumer line-up just not really cutting it for personal tastes anymore. Although I have a very hard time dealing with my HPX170, there are still several people that love the image and aren't as picky as us. They will continue to keep business, I won't be getting rid of my 35mm Adapter Rig by ANY means.

However, I can see myself having two to three of these VDSLR/EVIL's in my own arsenal, of the same brand preferably. So bad-ass the concept of low-profile acquisition with A-Grade results.

I just really hope the AVCHD @ 17mbps can really hold up to scrutiny. For some reason, with specs already published, I don't see any changes being made to the 1080i/60i 24P Capture nor the Bitrate... =T Surprises maybe?

 

[Park Edwards]

SX1 or GH-1? Neither...although I have been chatting with people who have it and looking at lots of clips. When I started I was looking at the EX3, EX1, HPX170, HMC150. Looking harder, none are interesting any longer.

However, the SX1 probably sets the low end of what should be considered acceptable. '

I have another GH-1 question: is it limited to 4GB FAT32 files or does it have a mechansim for continuous recording?

i read "could" instead of coulnd't be.

 

[LizaWitz]

That doesn't seem sensible. That's because the GH-1 is sharper than the MK II?

Dunno, but thats what I get when I plug in the numbers.

I have another GH-1 question: is it limited to 4GB FAT32 files or does it have a mechansim for continuous recording?

Recording time is "until the card's full" but, I figure it does like other cards and when it hits the 4GB filesize limite of FAT, it continues in another clip.

Oh, and why are we still dealing with the poor FAT filesystem? I remember in the 1990s there were several filesystems designed specifically for the needs of FLASH, which were open sourced. Too bad the flash manufacturers didn't have the guts to get together and standardize on a decent filesystem-- by now Apple and Microsoft would have adopted it (and if it was one of the open standard, Linux would already have it too.) Instead, for SCXC, they go with ..... FAT again. UGH.

I don't see any changes being made to the 1080i/60i 24P Capture nor the Bitrate... =T Surprises maybe?

Its 1080/24p or 720/60p in AVCHD or 720/30p in MJPEG. No interlaced option, thank god.

 

[Barry_Green]

Not accurate. The canon has a bigger
Sensor. You wouldn't use the same lens on both. They would have different FOV. The canon will always have shallower DOF.

Not sure this is right, but I tried running some DOF calculations. It looks like the GH1 offers shallower DOF than the 5D:

With a 40mm lens, f1.7, and subject distance of 10 feet:
Model: DOF, Range of acceptable sharpness in feet
5D: 1.91, 9.13-11
GH1: 0.95 9.55-10.5

40mm lens, f1.7, subject distance of 4 feet:
5D: 0.3, 3.86 - 4.15
GH1: 0.15, 3.93- 4.08

20mm lens, f1.7, subject of 10 feet:
5D: 8.94, 7.24-16.2
GH1: 3.96, 8.4-12.4

20mm lens, f1.7, subject 4 feet:
5D: 1.24, 3.47-4.71
GH1: 0.61, 3.72-4.33

Calculations from: http://www.dofmaster.com/dofjs.html

 

[Ian-T]

Its 1080/24p or 720/60p in AVCHD or 720/30p in MJPEG. No interlaced option, thank god.

I don't believe this is accurate either. According to specs the cam captures 24p in a 60i container. So I believe there is pulldown added. The 60p options are fully progressive though. (720p).

 

[Emanuel]

It does seem there are some mistakes and half trues which should be enlightened in behalf of the facts.

I don't think the Canon is throwing out 75% of the pixels. The interpixel gaps would be so huge as to have other artifacts. Also, it would lose its low light advantage, which is clearly not the case.



It does throw away 19% of the pixels, since 16x9 can only use 3159 rows.

The point is: when using (in the 5D Mark II's movie mode) every pixel on every third row -- AFAIK it ain't skipping every 2nd pixel (horizontal and vertical) but sampling only one out of 3 pixels (horizontal not vertical), the resolution drops to 1/3rd. Being the full resolution (sensor) in movie mode (before downsizing to 1920x1080) 5.9MP instead.

More info here:
http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

But the sensor size for DOF is still the FF35mm (36x24 mm). Therefore, the noticeable problems with aliasing or moiré pattern -- avoidable with filters during shooting or even decreasing sharpness, not so easy in post production though.

The differences ain't only reported to DOF. Here is something where you can easily understand the tech limitations of an acquisition device in any way other than as 35mm sensor size. Even though you have no noise reduction on high ISO going with the 5D Mark II in movie mode. Anyway, do not try above ISO 1600 (example below) or so. Or you shall not have a good surprise at all.

Courtesy by By John Mahoney

On the other hand, here is another example which can help to figure out what does mean an interlaced encoding for progressive recording:

Interlaced encoding can be a problem for the progressive modes, as it is possible to follow with the article brought by Adam Wilt:

http://provideocoalition.com/index.php/awilt/story/review_canon_vixia_hf11_avchd_camcorder/P1/

 

[Emanuel]

As conclusion, I'd easily admit that Canon has found a perfect balance or match between a higher bitrate and the necessary resolution in order to supply that well-known outcome we're used to check it out there.

 

[DavidNJ]

It is precisely the Nyquist theorems that have me questioning that. Do you have a source for your information on the the 5DM2 sample size?

There should be a low pass optical filter to reduce noise reflected above the Nyquist limit. That filter would be set to work with the sensors full 78 lp/mm sensor resolution. Reducing it by half would imply no optical filter at all. Wouldn't it?

 

[joe 1008]

The nanoFlash XDR is nice and most important portable, but expensive at $3500 and using MPEG2. Still 3x the bandwidth of a similar protocol XDCAM EX. And records I-frames at 160Mb/s.

Blackmagic Designs Intensity to a computer is $250-350 and Cineform is $500-750 (to capture directly). My guess is Cineform will also store a somewhat better image. To make Cineform portable you'd needs a battery powered PCIe to ExpressCard expansion chassis (I only know of a plug in model for $750 and a laptop with two disks and a fast processor...someplace between $1000 and $2000. Not that much different than the nanoFlash XDR. Then, they aren't shipping the nanoFlash XDR.

David, any thoughts about the Panasonic AG-HPG20 recorder? Should have a streetprice of about 4.500USD and is playing more or less in the same league like the nanoFlash - providing a probably even better codec.

Though at first sight it looks absurd to stack such an expensive unit to a 1.500 USD camcorder it's still in the EX1/EX3 range pricewise - even counting with a cage or some rails to keep it all together.

 

[DavidNJ]

Let me start by saying this is getting beyond my actual experience.

Some groundwork:

• AVC-Intra 50:
  • nominally 50 Mbit/s
  • CABAC entropy coding only.
  • 1920x1080 formats are High 10 Intra Profile, Level 4
  • 1280x720 formats are High 10 Intra Profile, Level 3.2
  • 4:2:0 chrominance sampling
  • frames are horizontally scaled by 3/4 (1920x1080 is scaled to 1440x1080. 1280x720 is scaled to 960x720)
• AVC-Intra 100:
  • nominally 100 Mbit/s
  • CAVLC entropy coding only.
  • All formats are High 4:2:2 Intra Profile, Level 4.1
  • 4:2:2 chrominance sampling
  • frames are not scaled

So, that compares MPEG-2 GOP at 100Mb/s or MPEG-2 I-frame only at 160Mb/s against MPEG-4 I-frame at 100Mb/s. At those bandwidths it isn't clear to me which would have the advantage.

The Panasonic is slightly more expensive but with probably better vendor support. Neither supports HDMI (the nanoFlash XDR does) so you would need a HDMI to HD-SDI convertor assuming the GH-1 output HDMI (fingers crossed).

Now, you could setup up a laptop to record a blackmagic card via magma expressbox, however you would need a battery power source to use it remotely. The cost would be less than $3k including the laptop.

 

[joe 1008]

I didn't consider that you need a converter. Thank you for that advise. But there seem to be compact solutions for a few hundred dollars. Though I would probably loose the compact form factor of the system. (one point for the NanoFlash)

One advantage of the AVCIntra codec is the 10bit encoding compared to 8bit of the NanoFlash. But would a possible HDMI of the GH1 output suport 10 bit? I suddenly got my doubts, did some research and found out that the HV20 has an 8 bit output and the EX1 gives full 10 bit.

So we don't only have to wait IF there is a HDMI output but also if it's 8 bit or 10 bit.

 

[DavidNJ]

There also the mapping of the signal...David Taylor's Cineform blog talks about some issue they dealt with.

One advantage of the PC solution is recording to Cineform 10-bit 4:2:2. The advantage of 10-bit storage of an 8-bit signal is for accuracy in editing. Is 2/3 1, 0.7, or 0.67?

 

[joe 1008]

The advantage of 10-bit storage of an 8-bit signal is for accuracy in editing. Is 2/3 1, 0.7, or 0.67?

Sorry, didn't understand this. 8 bit means 256 gradations for a color, 10 bit resolves 1024 gradations or about four times more information. This will give you much more lleway in post, don't expect any banding at all.

But one point that is very important for ME at least is to keep it simple AND compact. That rules out a laptop solution. Probably the NanoFlash would be the most compact and relyable quality solution? - after all I've learned for now...

 

[DavidNJ]

Quote:
Originally Posted by DavidNJ
The advantage of 10-bit storage of an 8-bit signal is for accuracy in editing. Is 2/3 1, 0.7, or 0.67?

Sorry, didn't understand this. 8 bit means 256 gradations for a color

In 8-bits the value is "2". In processing--color correction, transitions, etc.--the software divides the value by 3. In a base 10 analogy, with the same 8-places of precision the value could be 0 or 1. With one more digit the value could be 0.6 or 0.7. With two more values of precision it could be 0.66 or 0.67. In the end, this will yield fewer artifacts.

Sort of like the rounding error in Superman IV or Office Space.

 

[combatentropy]

10-bit log is far better than 8-bit linear, and even 12- or 14-bit linear. Because it is logarithmic, not linear, the shades are evenly distributed among f-stops. Meanwhile, linear analog-to-digital encoding wastes an enormous amount of bandwidth in the upper zones.

For more information read Human Vision and Tonal Levels, by Norman Koren.