Park Edwards
Veteran
so many wishful thinkers out there.
Firmware update V1.2 (2009-17009)
- Thread starter dobro
- Start date
so many wishful thinkers out there.
commanderspike
Carbonite Member
That was an entertaining read.
But I am sticking partially to my theory, that the reason the firmware update doesn't attempt to improve video quality is that Panasonic have a very clear spec laid out for each product line, and they cannot afford any overlaps.
They want consumers to buy a GH1. But not just a GH1. They want them to buy a camcorder as well. They must surely draw a line in the sand and say "Do not cross it".
Convergence is dangerous for profit margins.
They attempt to segment the market into as many pieces as possible so they can sell as many different units as possible, and they bring out new models in each line every year whether or not they actually have some new technology to show.
To buy a top end Panasonic camera and a top end Panasonic camcorder would cost a lot more than just buying ONE GH1.
Panasonic have proved with the GH1 that to (nearly) make the camera indie film makers dream about, does not cost a lot with modern technology. But I will bet my house on the pro version being marked up considerably. Because they can, and they will.
I find it unacceptable that these companies always have to make people pay the maximum of what they're willing to do, all the time. Panasonic could get a massive foothold in the indie filmmaker market here simply by selling a Pro GH1 at a reasonable price without the temptation to mark it up so much, they would get a much bigger reward in the long term and also enable lots of filmmaker community based 3rd party businesses to make a profit too, selling accessories and / or renting out cameras.
As for the technical reasons behind the GH1's poor codec, and mud... yes downsampling a 12 megapixel sensor to 1920x1080 is going to be taxing for the image processor, but why then can those horrible top end single CMOS chip sensored camcorders take 10 megapixel stills? Surely the must be downsampling to get to 1920x1080 quite a bit too, yet they cost less than the GH1 and have a better AVCHD implementation. What makes the GH1 'special' in this regard? Why technically couldn't they do it? I see no reason why mud could not have been avoided, especially given the time spent on 2 firmware updates.
But I am sticking partially to my theory, that the reason the firmware update doesn't attempt to improve video quality is that Panasonic have a very clear spec laid out for each product line, and they cannot afford any overlaps.
They want consumers to buy a GH1. But not just a GH1. They want them to buy a camcorder as well. They must surely draw a line in the sand and say "Do not cross it".
Convergence is dangerous for profit margins.
They attempt to segment the market into as many pieces as possible so they can sell as many different units as possible, and they bring out new models in each line every year whether or not they actually have some new technology to show.
To buy a top end Panasonic camera and a top end Panasonic camcorder would cost a lot more than just buying ONE GH1.
Panasonic have proved with the GH1 that to (nearly) make the camera indie film makers dream about, does not cost a lot with modern technology. But I will bet my house on the pro version being marked up considerably. Because they can, and they will.
I find it unacceptable that these companies always have to make people pay the maximum of what they're willing to do, all the time. Panasonic could get a massive foothold in the indie filmmaker market here simply by selling a Pro GH1 at a reasonable price without the temptation to mark it up so much, they would get a much bigger reward in the long term and also enable lots of filmmaker community based 3rd party businesses to make a profit too, selling accessories and / or renting out cameras.
As for the technical reasons behind the GH1's poor codec, and mud... yes downsampling a 12 megapixel sensor to 1920x1080 is going to be taxing for the image processor, but why then can those horrible top end single CMOS chip sensored camcorders take 10 megapixel stills? Surely the must be downsampling to get to 1920x1080 quite a bit too, yet they cost less than the GH1 and have a better AVCHD implementation. What makes the GH1 'special' in this regard? Why technically couldn't they do it? I see no reason why mud could not have been avoided, especially given the time spent on 2 firmware updates.
Last edited:
Ben_B
Veteran
Ahhhh here I was thinking in terms of processing power...I figured that if it was "wasting" processes doing B-Frames and leaving data hanging till the next frame and going back and dropping more data that would take a whole lot more processing power (which is of course the point of why the Gh1 probably doesn't have it.) But lets say it did....then why would I want to use more processing power for a smaller file size...in theory I would think that the opposite would happen: more processing power could be allocated to each individual frame in the group to be used for accuracy of the original directional interpretation or even to have more I-Frames in there, rather than just trying to save more space...maybe a pattern like IPPPIPPPIPPPIPPP? If B-frames really take so much more processing power why not use B instead of P and get more I frames?
commanderspike
Carbonite Member
Thats right Ben, about the B frames. As far as I know it, yeah they take up more processing power because the codec chip has to go back through already encoded footage and prop it up.
But I don't buy Barry's theory that the downsampling from a large sensor means the encoder chip cannot have time left to do B frames like a camcorder can.
First of all, whilst the camera is recording it outputs raw footage to the Venus HD image processor which then down-samples to approx. 460,000 pixels for the LCD screen and viewfinder. This raw stream is not affected by mud. So what happens between the LCD and the memory card?
Answer... the crippled AVCHD encoder.
Already the hard work with the sensor has been completed. The stream is then passed to the AVCHD encoding chip and the memory card.
The encoding chip has pretty much the same 24p raw HD output from the GH1's image processor to deal with as it would on a top end consumer camcorder. But yet makes a complete hash of it, and wraps it in the idiotic 1080i (for compatibility with idiots) before writing the MTS files to the memory card.
This smacks of a fishy artificial limitation or a poor, unnecessary mass market driven design decision. Why, if the sensor can output such a clean stream, down-sampled to the LCD, can the encoder chip not have time to do B frames like the same encoder chip in a Panasonic CMOS camcorder?
This is not about the CPU having a hard time with the large sensor, and it is not about down sampling.
But I don't buy Barry's theory that the downsampling from a large sensor means the encoder chip cannot have time left to do B frames like a camcorder can.
First of all, whilst the camera is recording it outputs raw footage to the Venus HD image processor which then down-samples to approx. 460,000 pixels for the LCD screen and viewfinder. This raw stream is not affected by mud. So what happens between the LCD and the memory card?
Answer... the crippled AVCHD encoder.
Already the hard work with the sensor has been completed. The stream is then passed to the AVCHD encoding chip and the memory card.
The encoding chip has pretty much the same 24p raw HD output from the GH1's image processor to deal with as it would on a top end consumer camcorder. But yet makes a complete hash of it, and wraps it in the idiotic 1080i (for compatibility with idiots) before writing the MTS files to the memory card.
This smacks of a fishy artificial limitation or a poor, unnecessary mass market driven design decision. Why, if the sensor can output such a clean stream, down-sampled to the LCD, can the encoder chip not have time to do B frames like the same encoder chip in a Panasonic CMOS camcorder?
This is not about the CPU having a hard time with the large sensor, and it is not about down sampling.
Barry_Green
Moderator
Okay, fine -- but here's what I'm really, really, really, really struggling with:
Why do people always think there's some grand protectionist theory going on with camcorder manufacturers, and reject every possible alternative explanation... but only have this attitude related to cameras? They seem to understand that a Mac Mini isn't "crippled" to keep it from competing with the Mac Pro. They seem to understand that a Hyundai Accent isn't "crippled" to keep it from competing with the Genesis. They seem to understand that a 13" Macbook Pro isn't "crippled" to keep it from undercutting sales of the 17" version.
Nobody thinks the Ford F-150 is purposely "crippled" to keep it from taking sales away from the F-350. In fact, the F-150 is the best-selling vehicle in America, obviously everyone there understands the price point vs. functionality relationship...
But when it comes to their video cameras... man, it's all about "crippling" and "boneheaded decisions" and "stupid executives" and "don't they get it" and "how dare they".
Why is that?
Why do people always think there's some grand protectionist theory going on with camcorder manufacturers, and reject every possible alternative explanation... but only have this attitude related to cameras? They seem to understand that a Mac Mini isn't "crippled" to keep it from competing with the Mac Pro. They seem to understand that a Hyundai Accent isn't "crippled" to keep it from competing with the Genesis. They seem to understand that a 13" Macbook Pro isn't "crippled" to keep it from undercutting sales of the 17" version.
Nobody thinks the Ford F-150 is purposely "crippled" to keep it from taking sales away from the F-350. In fact, the F-150 is the best-selling vehicle in America, obviously everyone there understands the price point vs. functionality relationship...
But when it comes to their video cameras... man, it's all about "crippling" and "boneheaded decisions" and "stupid executives" and "don't they get it" and "how dare they".
Why is that?
holdingpattern
Member
I'm not an expert, so someone can correct me, but my understanding is that B frames are predicted both from previous frames, and from frames that have yet to arrive (at least when viewed linearly). This means that the chip has to buffer several frames of presumably unencoded raw footage. The encoder then follows several frames behind looking for predictive behavior in the frames in front of it and behind it. So not only is there simplistically twice as many frames to analyze for similarities, but it has to buffer a lot of data and manage that buffer. Presuming they put just enough memory in the camera to process real-time without B-frames I can imagine there is a memory limitation in addition to a CPU limitation. In a relatively cheap camera, one of the easiest places to cut excess cost is on buffers that aren't needed most the time (i.e. for stills, which is what the camera is designed mainly for, it is a DSLR-like hybrid after all)
B-Frames help improve quality by creating a more efficient encoding of the footage. With a constant bit rate, this means more of the bandwidth can be used on the details in the image, and not sending potentially redundant data frame after frame.
Michael Olsen
Veteran
Because I want a sub-$10,000 camera that can actually resolve a real beautiful 1080p (if not ~2.5 or 3K), accept different lenses, offer the usual trappings of cameras (EVF, monitor, audio, levels, etc) and provide us with a rock steady codec. DSLRs are aggrivating because they seem "so close" to perfect. "If only" they would change a few things...
Which is why I think that these DSLRs are great, but that they aren't really RED killers. RED is setting out to break the mold from the outset. Nikon, Canon, and Panasonic seem to have just stumbled into the hole and are in the process of making sense of things. The images the cameras produce are good and depending on the user, great. But they all have noticeable problems in terms of rolling shutter, noise, aliasing, moire, macroblocking, etc. When someone takes the DSLR idea, develops them a little more (filter, sensor quality, acquisition format, usability, etc), and makes a real video camera with them...let the fun begin.
That said, I can't wait to start shooting! :beer:
Last edited:
commanderspike
Carbonite Member
That is the most convincing technical reason I've ever heard for why the GH1 doesn't handle B frames.
And I accept Barry's point about spec versus price point (although it's easier to accept Apple having a 13 inch laptop when they have a 17 inch option available. Panasonic don't have an option for a GH1 with a better encoding chip).
But what I still don't get is that a consumer camcorder can predict from the raw data B frames, and yet the GH1 - while it can output to the LCD a really nice stream, and output the similar stream to a MJPEG file, when it comes to AVCHD... oh...hell... No I think Holdingpattern is probably right on this.
It's not crippled, it's just technically not capable.
And yeah, RE: Michael Olsen, I agree on that... it's so frustrating to be so close to realising the full potential of the things the GH1 offers so amazingly at it's price point, and yet 2 firmware updates down the road be no closer and have no news of a follow up yet.
But then, of course if we knew this early what Panasonic were making next, all of us would probably wait and hold off buying a GH1. They have to allow it a shelf life in order to make a profit.
Barry_Green
Moderator
Ah, the definition of a free market economy. Charge what the market will bear.
Then you have a problem with capitalism? Which is okay, if you do, but -- in a free market, that's what companies do. It's pretty much at the core of the concept of a free market; build what you can, sell it for as much as you can.
What horrible top-end single CMOS chip camcorders?
Don't know. Only the engineers who built it will truly know. But I doubt the answer is "let's see if we can screw the customer even harder this time."
Perhaps that is because you are not a camera engineer, nor a codec engineer, nor a hardware engineer, and don't know the processing or chip power of these products? I don't know them either, and therefore I can't answer either. But I'm pretty sure by this point that it's not something that will be fixed by a firmware update. I agree with pailes that it's probably an inherent processing power limitation, dictated by the low price. Same reason a netbook has an Intel Atom processor instead of a Core 2 Duo.
Barry_Green
Moderator
Because more I frames would drive the codec bandwidth through the roof. I-frames don't use any long-GoP wizardry to shrink space. They're just straight easily-processed frames, like JPEG images. Which makes them perfect for editing, but not for squeezing down into the minimal space.
P-frames are (comparatively) easily generated. B-frames are apparently the big nasty super-intensive processor-powered workhorse, that uses up all the CPU time but also allows for tiny bitrates.
Barry_Green
Moderator
That's a great theory, sounds perfectly reasonable.
Barry_Green
Moderator
Like raise the price $8,500? If you're willing to pay $10,000 you can probably have most if not all of that stuff.
Does anyone keep in perspective that we're talking about a unit that costs $1500 with lens, or under $900 without lens?
IF YOU WANT IT CHEAP, COMPROMISES WILL BE MADE. Why is that so very hard for people to accept?
Agreed 1,000%. Actually 1,1100%. These products are not designed to do what people are trying to shoehorn them into doing (me included). The only difference, I guess, is that when I exceed its limitations, I'm blaming myself for trying to push it too hard. Everyone else seems to want to blame the evil, stupid, greedy, incompetent, bone-headed, protectionist, headphone-hating manufacturers for daring to give us something like a 7D or GH1 for under $2,000.
commanderspike
Carbonite Member
I don't hate Panasonic for giving me a GH1 :thumbup:
What is frustrating me though, is that the core technology is clearly there to make a absolute mega-dream-indie-filmmaker camera, but the market forces keep stop such a product from existing.
It's not like we have an option, other than to wait for a GH2, which may or may not be fixed or wait even longer for a pro camcorder with this core technology refined to the Nth but at a incredibly high mark up.
And yes... I do have a problem with capitalism, it has a lot to answer for. But don't worry, I'm not a Communist like that Obama guy.
*joke*
commanderspike
Carbonite Member
I guess this is the same reason the skin is peeling off my red GH1? Hmm. Good design costs nothing. The GH1 is so damn near perfect... a few dollars more on a memory buffer...
Barry_Green
Moderator
Wait -- I thought the whole broadcast lineup was the other option?
What I've been saying for months. They design the hardware the best they can to the price point they're given.
Of course. Because the idea of actually engineering, developing, prototyping, manufacturing, testing, refining, and finishing a new product takes no time at all. They could spring a GH12 on us, fully formed, today -- and give them away in cereal boxes if they really wanted to.
Sigh.
It's like those overweight people who -- they'll believe in fat pills, diet pills, water pills, weird machines that electrically stimulate their muscles, that weird vibrating band device to jiggle themselves, anything -- ANYTHING -- other than "eat less and exercise more."
These companies are doing the very best they can, right now. They are given a price target to hit, and they design the product to be the best it can be and to blow away the competition at that price point.
You should not expect a Red One at a $1500 price point. If you do, the blame lies with your mismanaged expectations, not with the product.
I've said before: if the GH1 costs $10,000 I'd throw it against the wall. But because it's only $1500 I think it's one of the most significant products to come down the pike since... well, probably since the HVX200.
commanderspike
Carbonite Member
Not unless I win the lottery, it isn't.
Really, is it too much to ask of Panasonic to just spend that little bit extra on stuff they know the enthusiasts and pros would lap up?
The amount of sales they lose to the 5D and are about to lose to the 7D is surely significant enough to think twice about the codec issue.
That Philip Bloom, who was a big driver behind GH1 sales (because of his genuine enthusiasm and footage on his blog) and whom shot some promotional footage for the UK press unveiling of the GH1 now says the GH1 isn't useable for pro work because of it's codec, would surely be enough to make Panasonic comment on the issue, at least... or explain why it's like this... I'd be interested to know.
To me, it makes GOOD business sense to get the product right, even if it is early days technology wise for this kind of product at this price, and it is the first in a new generation. Panasonic actually said they wanted to take their time over video and 'get it right', so no video mode in the G1.
So for them then to rush out the GH1 just a few months later with it's crap codec, didn't exactly do a BRILLIANT camera justice, did it? Despite what we can make of it, despite all the enjoyment, it is still a frustration of mine regarding mud and has been since the first horrifying encounter. I don't think my expectations of having a clean AVCHD implementation were unreasonable, given that the problem has never occurred before on any other AVCHD products.
The fact is the GH1 has the worst implementation of AVCHD of any camera, at any price point. Whether this is for technical reasons or otherwise, it sure as hell doesn't do justice to the sensor... that is kinda sad. But still - a very separate issue to my enjoyment of some of the great footage that comes out of the GH1. It's just that I want it to be better in situations where mud can occur. Or at least have the option or knowledge to hand of a fix, upgrade or future model...
Look at the first sunlight / tree shot in my video here 50 seconds in... the compression is all from GH1 mud and it kinda spoils the scene... http://www.vimeo.com/6525278
Last edited:
Ben_B
Veteran
Yeah so what I'm thinking is if a P-frame uses more processing power the camera is theoretically capable of recording using more I-frames and thus a higher bitrate, larger file sizes, etc, with the same hardware, resulting in a theoretical increase in image quality....that is assuming there isn't a memory buffer problem...which I don't see why their would be (seems like a long string of P frames would cause more problems than this..) so basically what it comes down to here is that they made decisions to handle the GoP in this way and while we might not necessarily like it or might think there is a better alternative (I would love 20 minutes of record time on my 4gb card instead of 30 for better quality image) it's what they decided to do for whatever reason, and that reason was probably not ignorance.
Barry_Green
Moderator
Of course not. We all want better, and we should all ask for better, and let them know what we want.
I just think that directing statements to the decision-makers is a far more productive use of time than complaining about nonexistent conspiracy theories, that's all. We're eating up tons of bandwidth and time on an issue that not one of us can do a thing about -- whereas telling your dealer, your sales rep, the people at the trade shows, etc. -- that will speak volumes, to those who need to hear it!
Are they losing sales to those two? The Canons have serious issues of their own. The 5D doesn't do a whole lot that the GH1 does, and costs 3x as much. The 7D looks more promising, but even then it still costs twice what the GH1 does, doesn't have autofocus, and also doesn't have B-frames.
Don't be surprised if spending a little more gets you closer to what you want, but if you're not spending a whole lot more, you're not going to get a whole lot closer.
None of these products are supposed to be used for pro work!
With that said, they will be, and the GH1 certainly is and will be, and so will the others.
And it's $1500 including lens. I've said over and over -- if it cost $10,000 I would throw it against the wall. It's only because of the price point that I think so highly of it.
I agree.
Agreed. The potential here is astounding. And, when the codec cooperates, the footage is absolutely unreal.
Agreed 100%. That's why, within two minutes of my first seeing a GH1 up at NAB, I went and cornered every Panasonic engineer I could find on the show floor and pretty much demanded that they build a broadcast version of it.
Last edited:
Barry_Green
Moderator
That is too bad... compression didn't do you any favors on those shots at all. I *hate* long-GoP compression...
Now, to be fair, that is *exactly* the kind of shot that we decided months ago that the GH1 shouldn't be used on. Wrong tool for the job. The GH1 simply doesn't like leaves/trees, especially with sunlight.
When HDV cameras first came out I went overboard trying to find areas where they'd fail, because I knew they *would* fail, and I figured people should know about the potential danger zones beforehand. And when rolling-shutter CMOS sensors first came out, I did the same thing, finding areas where they'd fail, because as a new and different technology, they *would* fail. Regardless of what names people call me, or whatever, I still think it's important that when these new technologies are foisted upon the consumer/end user, that it's vital that the user know what he/she can expect, and what s/he CAN'T expect, from the product.
HDV hates fire, smoke, strobing lights, rotating motion, splashing water. Rolling shutters hate fast horizontal movement, vibration, and magnetic fluorescent/HMI lighting. You have to know these things, or the camera will screw you over. In fact, DV has its problems, primarily with red or blue scenes; every month we get someone writing in saying "how come my club scene/dance hall shot/concert shot looks so blocky and awful" and when we look at the screen shot, guess what -- it's a frame from an all-red lighting scheme, or all-blue. A 4:1:1 codec cannot deal with a solid wall of red or blue, and that's something that you've just got to know.
The GH1's codec hates lots of motion and gobs of fine detail. Throw the background out of focus and it gets much, much better. The worst I've seen it look is when shooting a forest, especially in stark sunlight -- the sheer amount of detail overwhelms the codec and it pukes.
Basically you had a pretty much "perfect storm" for the GH1's codec to puke -- gobs of detail, green detail, with sunlight/contrast added, and lots of freeform motion.
Try that exact same steadicam shot in 1080p mode inside a house, and I bet you'll get a million times better results.