i dont believe its any malice on their part.
they probably took the fastest and least expensive route to release the camera on time
Af 105a announced in japan
- Thread starter tacotim
- Start date
they probably took the fastest and least expensive route to release the camera on time
Firstly, you are confusing lpph (lines per picture height) with lp (line pairs). If you did have 1360 photosites, you would be theoretically capable of resolving up to 680 LINE PAIRS. This is equivalent to 20kHz sampling resolving 10kHz audio - you need two samples to record each cycle. But in video terminology we don't refer so much to LINE PAIRS as LINES. A black-white pair is counted as two lines.
So, for a Nyquist limit of 680 lines, (340 line pairs) it only implies a fundamental sensor resolution of 680 resolution units in the vertical. At first sight, this seems to doubly reinforce the validity of what you say next......!
That would be true if, and only if, the sensor was fully read and fully deBayered. But the extreme likelihood is that it is not. It's not what happens in any DSLR or still camera for video, and the AF100 is a GH2 chip at core.
Think what happens in the case of the Canon sensor, used in the C100/300/500, and the basic theory of which Canon are quite open about. The used area is 3840x2160 in a Bayer pattern. But Canon treat it as a 1920x1080 pattern of 2x2 blocks, with each BLOCK forming a fundamental resolution unit. Hence a Nyquist limit of 1080lpph - not 2160. The other thing that follows from this is that each block represents R, G, and B, so you'd expect any aliasing over 1080 to be monochrome and not coloured. Is this starting to sound familiar.....?
If the AF100 used exactly the same technique, (2x2 block readout) it would imply a pattern of about 1210x680 resolution units, and if they were 2x2 a sensor pattern of about 2420x1360 in a Bayer pattern. But that is only 3,291,200 - which indeed falls well short of the count of around 12 megapixel. There is strong evidence that Panasonic are using a similar technique to Canon - but with a 4x4 basic resolution unit, not 2x2, and that would make engineering sense.
Put those numbers in, and it implies a basic Bayer sensor of 4840x2720 - or 13,164,800, far closer to what we know it must be!
In practice, the vertical count must be divisible by 9, the horizontal by 16, which limits the possibilities in reality. I've read one account which reckons it to be a basic res unit of 648, so a pattern of 1152x648, and therefore actual read sensor area of 4608x2592 - a total of 11,943,936. Frankly, whether it's 648, 657, 666 or 675 is relatively academic - that's the basic mechanism by which a 12 megapixel sensor can have such a Nyquist limit, and it explains why the aliases you do see are monochrome. The 4x4 basic resolution unit is fundamentally monochrome.
I think if you reread the above, you'll haveto admit a contradiction. The whole point about aliases is that they are inherent to false response above the Nyquist limit. If we are seeing them above 680 lpph (or 648, or 657 or whatever) that's proof that the camera Nyquist limit must BE that figure!! And above is described a mechanism whereby a 12 megapixel sensor can give such a figure.
If you see aliases, it's because the input detail was finer than the cameras Nyquist limit. That's the fundamental point. If you admit there are aliases above 680, you are agreeing that the Nyquist limit must be lower.
Which is why I remain 100% sure the fundamental resolution is somewhere around 648-680lpph. Sorry if I'm telling you things you'd rather not be hearing.
Practically, the camera is what it is, and many users are satisfied with it. I'm just giving the underlying theory.
Last edited:
I think this sums it up really well. The camera from a technical point, apart from the nobody knows why missing focus expand, is a joy to work with. But the images just don't deliver. Yes, I have seen good looking images from it, but I can't help but think you can only get such images in a very, very limited number of situations. Because I totally share the experience you mentioned here:
This is exactly what I found out with the camera.
Now saying that around here is guaranteed to have people jumping on your throat to defend their awesome camera they spent a pretty penny on. But it's just the truth. I'm not saying it's impossible to get good images. I'm saying that when you do get good images, it's a lot of work and just feels like the camera is not made for that and you have to work your butt off to compensate for what it lacks to just get a half way decent result. It's capable of decent images, beautiful if they are close ups in controlled lighting situations with very soft light. But that's about it. Unless you like artificially over sharpened footage from the M4/3 lenses that add artificial sharpness to your image. In the end of the day I just don't understand why people just don't accept the camera grossly underperforms but if you have the time and are willing to work double you can get it to deliver at least decent results. Because this is what it is. So why try to sugar coat it?
I know price gets thrown around a lot as an "excuse". But in my opinion, the AF100 is still a sub-par camera even at $3,000. If the new AF100A will cost street $3,000 without any rebates, then it may be alright. When you compare the AF100 with the GH2, this becomes even more clear.
Also, the whole concept that the AF100 has no competition at it's price point is just a malformed defensive opinion. The GH2 IS competing with the AF100 and it's cheaper. That it doesn't have all the features of the AF100 and still kicks it's butt and competes with it is just more credit to the GH2 and less to the AF100. A camera doesn't have to have the same features to compete with another. It only needs to match or surpass the results from such camera and the GH2 does that in spades. It has been a real eye opener since I started using both. Almost unbelievable.
I read a post, I think over at the GH2 boards, with somebody saying that he has replaced his AF100 with a GH2 for main camera, and it took a couple of months for him to come to terms that the cheap GH2 was just the better camera and now the AF100 sits gathering dust. I again share the same feelings. You will get way better looking footage with a hacked GH2 than you will with an AF100 9 times out of 10. So who cares if the AF100 is "technically better" when the images clearly aren't?
Two weeks ago I had the AF100 and the GH2 in a shoot and the GH2 was supposed to be only for stills and productions shots. It was a two day shoot and in the middle of the first day, I just rigged the GH2 up with a mattebox and shoulder rig and switched to it and never looked back for the whole shoot. Some of the people were wondering why the bigger camera was sitting by the catering cart while it was all being shot with a small photo camera. I just got sick of fighting the AF100 to do it's job!
Last Monday the director called me and said that the earlier shots looked much worse than the later ones and he wishes we had just used the same camera from the beginning. And this was the AF100 with a Hyperdeck. Not AVCHD.
It's just an undeniable fact.
Last edited:
I didn't say it was malice. It was tongue in cheek. But it has to be something else other than "the fastest and least expensive". I can't see that such a minor feature that even low end digital cameras have could cost so much or delay the process that much.
bottomline, they screwed up on that part.
doesnt matter 'why'
ill be selling my af100 soon and will be getting a BMC and a gh3. not because it sucks, but because it has done its job and has paid itself back. af100 lasted 2 years with me and it was great for its time. technology these days dont last that long anymore. im betting c300 being a great camera would probably last 1-2 more years before being eclipsed by a cheaper and newer technology.
I'm tecno-impaired, and don't understand most of the above (my education was all in film)... However, two weeks ago I attended a film festival where one of my documentaries, shot with the AF100, played along with another 30 minute film that was shot with an AF100 and a Canon DSLR (sorry, I don't know which, the cinematographer wasn't there to ask). The projection was via BluRay, and the setup was excellent, the venue seated 150. Both films looked really good. The other film had wide vistas and tracking shots in a forest that were obviously done with the Canon. Visually, in that environment, there was little to no difference in image quality - it took knowing the constraints of the two cameras to guess which was which - the majority of the audience couldn't tell - I should add that my documentary looked more like it originated on film because that's always been my goal.
I'm not implying that the AF couldn't be greatly improved. But given that, for many filmmakers, this is the best type of exhibition that they're likely to ever experience, resolution isn't the real issue. An improved MFT sensor with a wider dynamic range, better highlight and chroma handling, is really what's needed. Unfortunately none of the press on the 100a mentions an improved sensor. If it has one I'll buy it in a second.
I'm not implying that the AF couldn't be greatly improved. But given that, for many filmmakers, this is the best type of exhibition that they're likely to ever experience, resolution isn't the real issue. An improved MFT sensor with a wider dynamic range, better highlight and chroma handling, is really what's needed. Unfortunately none of the press on the 100a mentions an improved sensor. If it has one I'll buy it in a second.
Point taken. It's seems to me that folks are expecting a lot from a sub $5,000 camera. 90 percent of my videos end up on the web. The AF-100 is more than enough camera for that purpose. Due to massive losses I suspect Panasonic has dialed back their R&D, particularly for a niche camera like the AF-100, hence the somewhat underwhelming upgrade. Personally, I love this little camera and plan on using to the end of its' useful life. It may not be the latest/greatest but it gets the job done for my clients. Sure the AF 100 has some short comings, but I've never used a camera in my 40 year career that didn't. Others in this forum have moved on to more expensive cameras and I understand that. I hope their investments work out for them.
David W. Jones
Veteran
With the release of the low cost Blackmagic Cinema Camera, it tells me that within the next 2 years we will be seeing a huge change in the quality as well as pricing structure of modern cameras. Bottom line... Manufactures that don't adjust will get left behind.
Erik Naso
Veteran
You got a great deal!!!!I actually paid only about 15% more for a used F3 than what I paid for the AF100 so I got my moneys worth
Seeing as you've clearly done some research into sensor theory, it's disappointing to see you insist on spreading speculative misinformation as if it were an established fact. Your understanding of how aliases are created by the folding down of spatial frequencies above the Nyquist limit is correct. What's wrong is your assumption that on the AF100, that Nyquist limit is at 680 (where you see aliasing first start to contaminate the resolution chart). If that was actually the spatial frequency folding point, you'd see a field of aliases spreading downward from 680 into even lower resolutions.
What actually happens is that the spatial frequency folding point lies well beyond the AF100's 1080p resolution. It is the trailing edge of ultra-fine details that we see folded down as low as 680. Aliases don't appear at lower resolutions because the OLPF is strong enough to block them. The aliases gradually increase above 680 as the filtering strength of the OLPF diminishes. There's noting "fundamental" about 680 and there's still plenty of genuine image detail beyond it. 680 is merely the point where the trailing edge of the folded down aliasing field first becomes apparent.
DC
Cinematographiliac
Hmmmm, not quite the "AF250" I was hoping for, but I kinda dig this new AF100A model. The 10-bit output is very appealing to me, especially since I already own a Sound Devices PIX240. I wonder if the imaging sensor has been upgraded?
I may consider getting this camera. If anyone - right now - wants to buy my mint and gently-used AF100, make me an offer. :happy:
I may consider getting this camera. If anyone - right now - wants to buy my mint and gently-used AF100, make me an offer. :happy:
LOL… if you only knew. There is little speculative about what I say and it’s certainly not misinformation. But I’ll try to argue your points on a technical basis. All I ask is that you give it some consideration.
OK. I think the problem is how the chart is being perceived, and what the numbers on it really mean. It needs to be seen as a pure input device. At any given place, it inputs a certain spatial resolution into the camera, and this differs from place to place on the chart, yes?
The image pattern is the output, and how – by comparison with the original chart - conclusions can be drawn about what the camera is doing. What is key to appreciate is that aliasing generated by detail at any one location ONLY AFFECTS THE PATTERN AT THAT POINT.
A common misconception is that if some detail of (say) 1250lpph on the chart generated an alias (via folding back, and if the Nyquist figure was 1000) of 750lpph, it would affect the chart pattern at the position of the 750lpph ring. It wouldn’t. At the point on the chart of the 1250 detail, you’d see lines coarser than expected – because they were 750lpph! – but you wouldn’t see any problem or effect anywhere else.
At the 750lpph ring, you’d either see nice converging lines, of spacing the same as the original if it was below the Nyquist limit – or coarser ones diverging if above the limit. The point is that at that place, the pattern is only determined by how the camera responds to the spatial frequency there. Not by what happens anywhere else on the chart.
Let’s imagine that we’ve broken the chart down into sections, with parts to cover certain areas of interest in the chart I linked to. In particular, the converging near-horizontal lines around the 600-800 sector. We’ve built a complete chart consisting solely of near horizontal lines slowly converging left-right, with res of 720lpph on the left side and 750 lpph on the right. What do you think we’ll see on the image?
It will actually be similar – but the lines will DIVERGE from left-right, not converge as in the real chart. If you measured the line spacing, you’d actually find they were 640lpph on the left – 610lpph on the right. Hence the divergence. Point is that there are no frequencies other than between 720 and 750 to affect the result. It can’t be being affected by “ where the trailing edge of the folded down aliasing field first becomes apparent”.
Do the same thing with real charts of 610-640, 640-670, 670-700, 700-730 and you’ll see that the first ones match the left-right convergence of the real original, but at some point they turn into a left-right divergence.
Practically, a chart such as Adam used is comparable to looking at all these charts at the same time. And with the lines at angles other than near horizontal. It may not allow for such accurate measurements – but it’s far more convenient.
In the real world, this theory matches what people have found in practice. That the AF100 holds it’s weight reasonably well as a near-720p camera – but is lacking when up against a true 1080 camera. If anybody with an AF100 doubts it, try a simple test. Frame a shot with plenty of fine detail, then record it in 720p mode. Now switch to 1080 mode and record the same thing.
View the results on a 1080 monitor and tell me how much real difference you can see between the 720 and 1080 recordings. I’d be prepared to make a decent bet that it will not be very much. Do the same thing with a true 1080 camera like an EX1 or an HPX250 and the difference is night and day.
Last edited:
Jan_Crittenden
Veteran
No that attitude is not attitude, but rather a factual set f statements. Just because you want an Alexa for $5000 doesn't mean you can have one. With much restraint here, I restate that the features and benefits of the AF100 remain in a lead in its price point. It is a proud place to be from where I sit, as we designed a camera that does deliver the goods. You might want it to be more that what it is but what we delivered is a fine camera and it continues to be. I am fairly proud of it and thankful to the group of DPs, Cinematographers and Film School faculty as well as some ASC members that helped design it. Are there concessions? Yes there are some things we could not do and meet the price point. So while that might sound like a price argument, compare feature for feature on the cameras against the other cameras at the same price point and you will see what I mean.
Best,
Jan
DC
Cinematographiliac
Thanks, Jan, for adding to this thread.
As I mentioned in my previous post, I may be interested in an AF100A.
A couple of questions...
1. Is the imaging sensor upgraded at all? Or is it the same one?
2. Will there be any kind of upgrade path for current owners, be it a trade-in or modification of the current model?
Kind regards...
Steve Kahn
Well-known member
Have you seen the BMCC? That Is very close to Alexa for 3K. Those were the features we want.
You can add all the features you want. You can Add a blender feature so that it automatically makes margaritas after shoot but if we don't want it we just don't want it.
As always I just get the feeling that Panasonic doesn't listen to its customer just like Canon didn't listen when they offered the 5Dmk2. At the time shooters and DP's around the world were fighting to get shallow dof with those terrible 35mm screens when Canon launched that camera. Was it intended to finally take on Red? What is it intended to help out all of those DPS who wanted a film look? Was it to revolutionize the way people look at the dSLR's
No. Canon was so blind and so oblivious that they enabled video recording on the camera at 30 frames per second for photojournalists who might also want to record video to accompany their stills for their new stories primarily to be shown on the web. 30p.
Panasonic today is just as blind to the needs of their customers . But you guys don't seem to be as lucky as Canon. They stumbled upon something great. They just got lucky.
electricpig
Well-known member
Asking as a non-science person, is this spatial harmonics, and hence similar to an understanding of sound waves and flow of objects through water and air?
electricpig
Well-known member
I don't understand this obsession with absolute resolution.
Images are getting too sharp on screen , too much detail, and certainly a faked looking sharpness and noise-free image over the look of film.
And neg film, once you've got that beautiful film grain dancing around on screen, gains an organic softness and dynamic range/falloff that the Alexa and BMCC deliver in spades, and exposing like transparency film (ie tight exposure range before things clip) is something the AF100 delivers if the user is prepared to learn and feel.
At the end of the day everyone is trying to emulate film with these large sensor cameras, but then want a super sharp noise-free image - go figure.
Images are getting too sharp on screen , too much detail, and certainly a faked looking sharpness and noise-free image over the look of film.
And neg film, once you've got that beautiful film grain dancing around on screen, gains an organic softness and dynamic range/falloff that the Alexa and BMCC deliver in spades, and exposing like transparency film (ie tight exposure range before things clip) is something the AF100 delivers if the user is prepared to learn and feel.
At the end of the day everyone is trying to emulate film with these large sensor cameras, but then want a super sharp noise-free image - go figure.
To be quite honest, I don't think Panasonic are blind to their customers needs - just strapped for R&D cash.
From a technical viewpoint, it's obvious what they need to do - develop a large-format sensor that is optimised for video. Not just make the best of one lifted from a stills camera. (Even though they may have made the best of it.) It wouldn't even matter whether it was 4/3 or s35 - just do a similar trick to Canon and Sony, make it 3840x2160 with 2x2 read out, whatever the overall size.
But from everything I hear, sensor development is an expensive business.......
But if they don't do some such development soon, Canon, Sony, and Red will corner the large format market in the higher end price bracket. Is Panasonic really happy to just let that happen? Just be a player in the cheaper end of the market - not matter how good a value for money it may or may not be? Just where is the Panasonic competition to the F55, the F5, the C300 etc?
electricpig
Well-known member
Or you could say that they were lucky enough for people to use the built in 'movie' mode, and a lovely dynamic range to make some lovely looking footage when things were perfect, but that failed on such fundamental video technical basics such as aliasing, moire and compression issues.
I just cannot believe these issues get past broadcast standards, and despite them have enabled the 5DII and the like to gain such a credible foothold.
I personally charge a small premium now if I have to do any serious post with 5d footage as it's is such all-consuming black hole to piss your time into to fix.
Before you then start the job you've actually quoted the time for.