AF100 / AF101 - initial report

you will see little "visual" difference, it's when you really start working with it you see it. Color Grading is one, huge difference in how you can push it around in the grade. One can argue get it all you can in camera as you want, that should be done anyway, but when you really need to save a shot, 8 bit vs 10 bit, you'll see why 10 bit is so important.

Yes VFX work, every last bit of information of the picture could mean that much cleaner of a composite or motion tracking.

Tied into that, 10 bit also means less generational loss the more you transcode it. Exporting out from your NLE, into your effects software, rencoding and encoding, stacking more and more on it, just to reimport it back in your NLE, and than exporting it out back again for your delivery, you'll see how much 10 bit can maintain it's quality more than 8 bit.

I'm not saying 8 bit can't look as good as 10 bit, that's not true at all. It's when you really need to dig into it, and put it through the vigors of a intense workflow, 8 bit starts falling apart on you. As I do colorist work myself, I want to be able to help and make sure the image looks the way the DP and Director wants it to look like when they couldn't get it in camera, for whatever circumstance. The worst thing for me to tell a client is, I can't do it, the footage falls apart when I try.

Same with VFX artists, and so on.
 
*Bump*

I think my untimely post may have gotten lost in a maelstrom of 8-bit vs 10-discussion. So, i will humbly try again:

Certainly a moot point, but I am wondering if someone could explain on a layman's level why the AF100 uses a CMOS chip technology rather than CCD. Is it simply a cost issue; 4/3" CCD's could not meet the sub-$6k price target? Or are there other reasons why CMOS made more sense? I would think CCD's would have better low light sensitivity and be immune to the rolling shutter issue.

Barry? Jan? Anyone? Thanks!
 
tangent said:
Certainly a moot point, but I am wondering if someone could explain on a layman's level why the AF100 uses a CMOS chip technology rather than CCD. Is it simply a cost issue; 4/3" CCD's could not meet the sub-$6k price target? Or are there other reasons why CMOS made more sense? I would think CCD's would have better low light sensitivity and be immune to the rolling shutter issue.
Click to expand...

Here is simple explanation.
Large CCD consume more power and heat like hell in constant readout mode.
Plus their production cost is quite high in last few years (despite common myth CMOS is not cheaper by design, but production facilities is aimed to CMOS for a reason).
Main problem with CMOS imagers had been noise, and for Panasonic especially (as they use slightly different NMOS technology their sensors are especially prone to fixed noise). But solutions for this problems become better and better.
So, low light sensetivity with same SNR for CCD in same price range is not better.
Yes, global shutter is only future, and is not possible to implement today in widely produced and cheap CMOS imagers. But we'll see it quite soon. As well as ultrafast contrast AF.
I know other reasons against CCD, but they require more detailed explanation.
 
r&d for large size CCD sensors also takes more time and is more expensive than pushing through a CMOS research pipeline, the production silicon line available to make CMOS is much more cost effective, and that's why you see CMOS sensors making leaps and bounds in improvement over CCD.

sensitivity is getting better, and noise floors are better off cmos censors now than any CCD camera. Heat is also a direct cause of noise, and the fact tha CMOS can run cooler at larger sizes too makes it a win, also when you want really high speed photography, CMOS is essential because of all these areas. Shooting 1000 fps you'll need a sensor easier to cool down.

Global shutters are very expensive.
 
tangent said:
*Bump*

I think my untimely post may have gotten lost in a maelstrom of 8-bit vs 10-discussion. So, i will humbly try again:

Certainly a moot point, but I am wondering if someone could explain on a layman's level why the AF100 uses a CMOS chip technology rather than CCD. Is it simply a cost issue; 4/3" CCD's could not meet the sub-$6k price target? Or are there other reasons why CMOS made more sense? I would think CCD's would have better low light sensitivity and be immune to the rolling shutter issue.

Barry? Jan? Anyone? Thanks!
Click to expand...
The Panavision Genesis, Sony F35, and Dalsa Origin are the types of video cameras with single-chip 35mm size CCDs. Those are cameras that cost hundreds of thousands of dollars, if they are even available for purchase.

And they are massive cameras to deal with the heat of their CCDs.

A video camera the size of an AF100 with a 4/3" CCD would likely melt.
 
CCD requires extremely high grade silicon that is probably a very low percentage of what gets sifted through in the factories. It takes a long time to get the amount of silicon needed, and to get that grade of it. You wonder why a 1/3 size, 3 ccd camera can cost you 5 grand? and why the dvx100 doesn't drop to 500 dollars? One of the major reasons.
 
xmephestox said:
CCD requires extremely high grade silicon that is probably a very low percentage of what gets sifted through in the factories. It takes a long time to get the amount of silicon needed, and to get that grade of it. You wonder why a 1/3 size, 3 ccd camera can cost you 5 grand? and why the dvx100 doesn't drop to 500 dollars? One of the major reasons.
Click to expand...

Where do you get this information? Being a CMOS chip designer myself, I've NEVER seen any issue with the quality of the silicon in any of the design parameters ever. In fact, the biggest issue with both CMOS and CCD design is process variance, which is the chemical/physical changes you make to the silicon with the addition, doping, etching, etc of that substrate. There can be a lot of variance in how you dope the silicon, how you grow the gate thickness. How you deposit the interconnect and vias.

CCD is expensive because it's a custom process. Every time you try and make changes to CCD technology, there isn't a LARGE volume of experience or in cost effective processes that you can start with. CMOS is used EVERYWHERE in chip design, so you have the opportunity to use multiple existing high volume foundries and with using modeled silicon that has been tested in REAL silicon ten fold to give you much more accurate simulation (and therefore tolerant design practices) AND allow the foundry to refine their processes.

Further, CCD technology doesn't allow (as easily) digital circuitry to be as easily integrated into the same die. That is the beauty of CMOS, you can put a lot of digital circuitry to drive/read the analog circuitry right there together on the same system (although this also brings up the issue of adding noise to the analog side of the design which IS a concern). Which ultimately reduces cost.

Global Shutter is not something inherent only to CCD. You can create CCD without global shutter and you can create CMOS with it. The cost per say is not in dollars, but in performance as well. Adding more control circuitry to the pixels removes their active area AND adds more noise potential. With CMOS you are limited with how you can control that global shutter as the basic operating layers of the silicon design are already predefined by the foundaries (or the modifications you can make).

At the end of the day, making a CMOS sensor is cheaper because you are working in a subset of design characteristics and layers that someone else has limited you with so that you can work in a known set of parameters.

Given the variance in CMOS through process and temperature, I will say I am amazed at how well CMOS sensors work. There is a LOT of variance in these processes, ESPECIALLY relative to temperature which is always an unknown. Analog design is always a huge challenge and my hats are off to ALL the latest CMOS sensors lately. They are very very impressive.
 
In further reading, it looks like with today's foundaries, the cost is relatively the same at the sensor level, but not so much when you have to push all the control circuitry for CCD somewhere else in the system (like other CMOS circuits and such). Have a look at this:

http://www.dalsa.com/public/corp/Ph...sg=AFQjCNH_MAx8RFYXoT2yrePw94dGgO11Cg&cad=rja

It looks like they do use modified foundry parameters for CMOS which probably explains why we get a lot more accurate devices in todays CMOS sensors. Thanks xmephestox for the pointer to where there was more info. :)
 
sorry for the confusion, i was reading in another article that cmos can be built under a more common silicon production line, while reading in another of the vigorous quality control that needs to go into creating silicon, wires got crossed somewhere in there, and i'm in no position to argue. CMOS does have the most potential simply in the fact that it's more widely used from the consumer to the professional, which allows for more broad research in it, and as stated before, a cooler running sensor that eats up less power is a huge advantage.
 
In any case, thanks for getting me to look at that Dalsa information, it was really good and can help people understand some of the differences of why CCD vs CMOS and a little bit of the details involved in both systems.

I'm just excited that within the last year CMOS imaging technology available to the consumer has really progressed far! I hope it continues down this trend!

Can't wait to pixel peep the AF10X :)
 
Is this camera going to utilize the same batteries panny uses for the hpx 170,hmc 150, hvx 200 etc. practically all their handheld cameras? Or will I have to purchase a new line of batteries and chargers?
 
monkeyking said:
Well, the Lumix 7-14 f/4 is finally back in stock. That's a 14-28 in FF speak . No IS though. Sharp and wide.
Click to expand...

I just ordered one. I weighted it up agianst the Olympus 9-18mm but the 7-14 appears sharper (probably helped by the Panasonic electronic integration) and it's obviously wider.
 
ripupthehwy said:
Is this camera going to utilize the same batteries panny uses for the hpx 170,hmc 150, hvx 200 etc. practically all their handheld cameras? Or will I have to purchase a new line of batteries and chargers?
Click to expand...

Same battaries Panny uses for HMC only. Not of HPX or DVX.
All are Handheld, but AVCCAM batteries are different than others.
 
Barry_Green said:
This is entirely true, and there are computers (and NLEs) for whom native AVCHD is a beastly burden to put up with. But on CS5 with the Mercury engine, they're pretty much seamless and I drag and drop the files onto the timeline all the time.

That said, it's not like P2, where P2 is a high-speed edit medium. Playback is chunky and slow off the SDHC cards and performs much better when copied to the hard disk. With P2, I can easily get five or six streams of HD simultaneously off the P2 card, but with SD it's not that way.
Click to expand...

I also noticed the difference between P2 and AVCHD transcoded to prores in FCS:
if I do quick edits on my powerbook I can layer more than 4 video streams in FCS
as soon as I hit one of the AVCHD files the computer gets a hiccup!

P2 is so incredibly smooth and NLE friendly its pure joy to work with!!

So Barry: can one record to the SD cards and simultaneously have the outputs from HDMI and SDI and record via those?
 
Jan_Crittenden said:
Please follow the guideline posted in the FCP white paper on the www.panasonic.com/AVCCAM site.
Click to expand...

Hi Jan,

if I go to that url
I find
AVCHD to P2 DVCPRO HD Transcoder

but then it only has the AVCIntra decoder ...

If there would be a transcoder that turns our AVCHD code into pure P2 format I would like to try it ;-)
The ProRes 422 QT transcoding is ok but does not behave as clean P2 material ( in FCS )
 
Barry, is there any possibility of any new Metadata feature until the release?

I am so waiting for some practically working Metadata features that can be useful for Filmmaking.
 
Last edited:
You must log in or register to reply here.
Back
Top