Af 105a announced in japan

There's an ecosystem Panasonic risks loosing if they can't maintain customers. Most AF owners likely own GH cameras, & some m4/3 lenses. And GH owners who've adopted the AF partially because they have the glass.

m4/3 is attractive in no small part because of it's commonalities. Loose one part, say the AF, & the rest becomes much less attractive. If I move to a Canon Cinema camera I'm much more likely to use my 5D3 in place of a GH3, & certainly won't buy any more Lumix lenses.

I'm sure Panasonic is aware of this, & hope they have a solution.
 
Last edited:
TheDingo said:
...I think many of us are saying that we are going to switch to one of Panasonic's competitors because Panasonic isn't producing a product that is viable for us.
.
Click to expand...
Exactly.

And the anger is from what we see as planned obsolesamce on the part of Panasonic. A lack of technology never prevented 10bit out when the AF100 was released 2 years ago. And a wave of vicious dedicated research by Panasonic's crack team of scientists and engineers didn't make it so that 10bit out would finally be available today.

Worse the "a" model will be obsolete before it is even launched.
 
Last edited:
Bern Caughey said:
There's an ecosystem Panasonic risks loosing if they can't maintain customers. Most AF owners likely own GH cameras, & some m4/3 lenses. And GH owners who've adopted the AF partially because they have the glass.

m4/3 is attractive in no small part because of it's commonalities. Loose one part, say the AF, & the rest becomes much less attractive. If I move to a Canon Cinema camera I'm much more likely to use my 5D3 in place of a GH3, & certainly won't buy any more Lumix lenses.

I'm sure Panasonic is aware of this, & hope they have a solution.
Click to expand...

This is so true - I have a GH2 & GF2 as well as my two AF-102s and I was planning to grab a GH3 when it's released. I've been happy enough with the AF cameras (but not absolutely thrilled) but I am starting to loose work to people with newer/better cameras (it's a fickle market), so I'm planning to update in the next six months.
At the moment Panasonic hasn't given me anywhere to go - the A doesn't really cut it, so there is a good chance I'll be splashing my cash with one of the competitors (even though it's going to cost me more)
 
lpowell said:
These last two paragraphs really bring your verbose handwaving into focus. What's clear that you haven't actually tried the 720p versus 1080p comparison that you suggested on the AF100.
Click to expand...
Without any verbosity, I can tell you I have indeed tried exactly what I suggested. Not only tried it, but did it with someone else who controlled the camera settings. Hence when I was judging, it was without being aware which was which. And the same with comparing an EX3 in 1080 and 720 modes.

Result was that for the EX3, viewed on a 1920x1080 monitor, it was night and day between the 720/1080 modes. For the AF100, there was not a lot of difference. (The 1080 mode did look a little better, which I would put down greater smoothing through extra upscaling - not because it contained any more actual detail.)

But please, I don't want anyone to simply take anything I say on trust, to just believe me over lpowell because of my "verbose handwaving". If you have an AF100 - try it. All you need is an AF100, an HDMI cable, and a good 1920x1080 screen. Then try switching the camera back and forth between 720 and 1080 modes and see how much (or little) difference there is. If you can, repeat the test with a camera with 1920x1080 chips (EX3, HPX250 etc). See the difference?

Did you really think I'd suggest an exercise which readers can so easily duplicate, if I had not already tried it and was sure of the results?
Your dogged insistance that the AF100 is a "near-720p camera" is demolished by this real-life comparison with the GH2:

http://www.dvxuser.com/V6/showthrea...-japan&p=1986231108&viewfull=1#post1986231108
Click to expand...
I don't see how that demolishes it at all? The GH2 is supposed to have a resolution about 50lpph more than the AF100, which puts it about 700-730. The lower of those two photos is somewhat sharper than the upper (eg detail on the fencing) so it's exactly what I'd expect.
Take careful note of the qualifying adjectives used in all three reviews: not maximum resolution, but "clean", "alias-free" resolution. What they're referring to is the perceptual point on the test chart where aliasing first starts to appear in the image. They are not claiming that the camera records no perceptible details beyond 680p, only that the details it records beyond that point are not entirely "clean" - i.e. they contain aliasing artifacts in addition to genuine image details.
Click to expand...
I'll try one more time to explain. You don't ever get "aliasing artifacts in addition to genuine image details" - at least, not on test charts, that's the whole point. Point a camera at a set of parallel lines and you will either get a true representation of what the camera sees - or a false one. If it's true, it's because the spatial detail is below the Nyquist limit - if it's false it's because it's above. You will either get genuine image details *OR* aliasing artifacts if you present a single spatial frequency at a time.

So imagine you had a set of charts, each with horizontal lines, and such that each one corresponded to a given spatial frequency. At first, with coarse lines, the image would be identical to the output. Same number of lines on image as on chart. As you got above the Nyquist limit then the image becomes "false". It may not look so supercifially, but if you accurately counted you'd find there were less lines on the image than on the real chart. The principle is EXACTLY comparable with what I said earlier about audio aliasing. It's comparable to putting single frequency tones into a 20kHz sampling system of 7.5,8.5,9.5, 10.5, 11.5 etc kHz. You'd find the respective outputs would be 7.5,8.5,9.5,9.5,8.5 etc - the last two being aliases.

So - if we did look at a chart of SOLELY lines of (say) 740lpph, and found the image was lines at 620lpph, how can that be "aliasing artifacts in addition to genuine image details"? There is no output at 740lpph (so no genuine image detail) and no input at any higher than 740 frequency, so no such to give aliasing artifacts. The "false image" you see is purely down to how the camera responds to detail of 740lpph.

Now for audio, you could use a frequency sweep instead of discrete frequencies. At the input, you'd see a continuously rising tone of (say) 1kHz to 19kHz. At the output you'd see a smooth rise from 1-10kHz, then it ramp down from 10-1kHz. That's aliasing above the 10kHz Nyquist limit.

So wouldn't it be great if we could do a similar thing in the image domain? And that's exactly what such as zone plates, and such as Adams chart are designed to do. Concentric circles getting closer and closer together in the case of zone plates, lines converging towards the centre in the latter case. And the output is how the camera image of the chart appears. Instead of a tone rising, then falling as you pass the Nyquist limit, here you see lines converging, then diverging as you pass the Nyquist limit.

Which is why the point at which the lines stop converging, and start diverging on the image is so important in analysis. It's the Nyquist limit for the sensor/sampling system. And from Adams chart, for the AF100, it's at around 680lpph.
 
noirist said:
harddrive, thank you for your careful well-reasoned postings and for maintaining your equilibrium in the face of inflammatory language and ad-homins.

Your proposed experiment only holds if the 720p mode on the AF100 produces a measured 720p. Some cameras produce less than 720p in 720p mode, and the AF100 could as well.
Click to expand...
Firstly, thank you for the compliments! :)

But I have to disagree with the next statement. (Sorry!) The reasoning goes back to post #182, where I tried to explain a mechanism whereby a sensor of about 12 megapixel may have a far lower actual resolution in actual use. The principle is far from just theoretical - the Canon sensor has a photosite count of 3840x2160, but only 1920x1080 resolution when read out for video.

That's forming pixels from 2x2 blocks. The AF100 seems to be doing similar with 4x4 blocks. Hence (using the 680 figure) it seems likely that the first thing derived is a grid of about 1210x680 pixels (RGB). That then either gets upscaled to a 720 or a 1080 raster as desired. It follows that the real resolution in each case will still be 1210x680. Hence, if you switch between 720/1080 modes and don't see a lot of difference, it follows that the underlying camera resolution MUST be less than 1280x720. See a big diference and the camera is capable of better than 1280x720.

That said, 1080 mode will (and does) appear somewhat better, regardless. It's akin to taking a low res photo in Photoshop and upscaling. It looks "smoother" - even if there is no extra real resolution.

One point about price. Yes, it's significant, and no argument when Jan says she believes Panasonic are giving value for money.

But as others have pointed out - sole references to basic body cost are only part of the story. Lenses and other accessories can easily cost more than the basic body. In which case, what might seem a decent saving on body cost, can seem a far smaller saving when looked at as system cost.
 
Last edited:
harddrive said:
You don't ever get "aliasing artifacts in addition to genuine image details" - at least, not on test charts, that's the whole point. Point a camera at a set of parallel lines and you will either get a true representation of what the camera sees - or a false one. If it's true, it's because the spatial detail is below the Nyquist limit - if it's false it's because it's above. You will either get genuine image details *OR* aliasing artifacts if you present a single spatial frequency at a time.
Click to expand...
Sorry, it doesn't matter how many ways you try to explain it, your assumptions are just plain wrong. For an example of alias artifacts superimposed on top of genuine image details, examine the Zone Plate in Figure 2 in Alan Roberts' AF100 BBC report:

http://thebrownings.name/WHP034/pdf/WHP034-ADD66_rev_2_Panasonic_AF101.pdf

Expanding from the center of the chart, you'll see dark concentric circles of genuine image detail. From the horizontal and vertical perimeters of the disc, you'll see light gray semi-circles of alias artifacts radiating toward the center of the chart. Those aliased circles have been folded down into the visible resolution range from originating Nyquist points that lie well beyond the 1080p resolution of the image sensor.

In the midrange areas of the chart, you'll see light gray cross-hatched patterns where the genuine image details radiating from the center of the chart are combined with aliased artifact circles radiating from the perimeters of the chart. The inner circle where the aliased circles first start to become visible is the 680p point that Roberts judges as the boundary of the sensor's "clean" resolution. In perceptual terms, there's nothing special about that point and there's still plenty of resolution beyond it, comprised of both genuine image details as well as aliased artifacts.

The reason these artifacts are called "aliases" is because it's usually difficult to distinguish them from genuine image details. In this case, the only reason we can spot the aliases in this Zone Plate is because we have advance knowledge of what the chart ought to look like. If it was a real-life object that was not contrived with geometric patterns, we'd have no way of separating the genuine details from the aliases. That is a major reason why digital images look realistic, in spite of being loaded with subtle alias artifacts.
 
Steve Kahn said:
Exactly.

And the anger is from what we see as planned obsolesamce on the part of Panasonic. A lack of technology never prevented 10bit out when the AF100 was released 2 years ago. And a wave of vicious dedicated research by Panasonic's crack team of scientists and engineers didn't make it so that 10bit out would finally be available today.
...
Click to expand...

Do the design, engineering, marketing, and management teams at Panasonic know you've been privy to their strategy sessions?
 
Steve Kahn said:
Exactly.

And the anger is from what we see as planned obsolesamce on the part of Panasonic. A lack of technology never prevented 10bit out when the AF100 was released 2 years ago. And a wave of vicious dedicated research by Panasonic's crack team of scientists and engineers didn't make it so that 10bit out would finally be available today.

Worse the "a" model will be obsolete before it is even launched.
Click to expand...
Overstatement. People still shoot with hvx200, people shoot 7d or dslrs. Af100 has its own quirks but definitely usable for certain work. Its not a cam for evetything.

Re: resolution argument
Ive compared dslrs with af100 and the af100 do resolve more than them albeit less pronounced due to the olpf.
 
lpowell said:
Sorry, it doesn't matter how many ways you try to explain it, your assumptions are just plain wrong. For an example of alias artifacts superimposed on top of genuine image details, examine the Zone Plate in Figure 2 in Alan Roberts' AF100 BBC report:

http://thebrownings.name/WHP034/pdf/WHP034-ADD66_rev_2_Panasonic_AF101.pdf

Expanding from the center of the chart, you'll see dark concentric circles of genuine image detail. From the horizontal and vertical perimeters of the disc, you'll see light gray semi-circles of alias artifacts radiating toward the center of the chart. Those aliased circles have been folded down into the visible resolution range from originating Nyquist points that lie well beyond the 1080p resolution of the image sensor.
Click to expand...
No, I'm not wrong, and what is frustrating to me is that I know exactly why you are thinking as you do. It is exactly the way I looked at the issue when I first started dealing with such charts - so I understand exactly where you're coming from.

What you are saying is that the "distortions" you see at lower spatial frequencies are due to aliasing from the higher spatial frequencies, yes? In simple words, that the artifacts you see on inner circles, are caused by aliases generated by outer circles? I believe that's what you mean by "Those aliased circles have been folded down into the visible resolution range from originating Nyquist points that lie well beyond the 1080p resolution of the image sensor."?

Just think what would happen if you covered up nearly all the chart with white card except for a small square somewhere in the region where you say you see "genuine image details radiating from the center of the chart are combined with aliased artifact circles". Now if your argument is correct, the outer rings, the only source of very fine detail, are now covered. With the fine detail gone away, presumably you would argue that the aliases would go away, and you are now left in this square with just genuine image detail corresponding to the original chart?

That doesn't happen. No matter how small a square you leave, the pattern in the image won't change at that point - at all, compared to when the entire chart is visible. And yes, I have done this for real as a demonstration. You are ONLY now putting a small range of spatial frequency into the camera, so any distortions from the original chart must be due to that frequency alone.

It took me a while, but my lightbulb moment came when I realised that the lpph detail indications *ONLY REFER TO INPUT*. I used to think that a more outer ring (say 1250 lpph), would, if it aliased to 750lpph, give an artifact at the radius on the chart corresponding to 750lpph - and I think this is what you are saying. It's not true - it will give it's alias at the same place on the image (on the 1250 radius), but the difference will be that the lines will be broader there than on the real chart. They will be 750lpph, rather than 1250lpph. And because of the geometry of the chart - the ever closer rings as you go out, and the circles themselves - the aliases show up as "ghost" circles centred on a different point. (Which is actually twice the Nyquist limit.)

If dealing with horizontal and vertical lines is complicated, then diagonal lines (and arcs of a circle) gets even worse. You have to think in terms of vector components. For a straightforward Bayer sensor, it will have a higher resolution to diagonal lines than hor/vert which is why you start to see "genuine resolution" beyond the Nyquist limit as you move onto the diagonals. This is what becomes visually misleading on zone plates.

Honest, lpowell - I do know what I'm talking about. Give me a day or two and I'll try and find a definitive reference.
If it was a real-life object that was not contrived with geometric patterns, we'd have no way of separating the genuine details from the aliases. That is a major reason why digital images look realistic, in spite of being loaded with subtle alias artifacts.
Click to expand...
To an extent that is true. But there is a big difference between stills and video. With stills, the aliases just degrade the image slightly - with video they can cause a disturbing effect on such as moving sharp edges. Maybe not too bad just to the eye, but it can cause problems with video compression systems. So the camera footage seems reasonably OK, but degrades heavily on transmission.
 
Igelkott Film said:
Here's a list for all camcorders you can buy on BH-photo that supports 10-bit.
Click to expand...

This is an internal recording 10-bit list, which the AF100A would not qualify for.

If you switch to a 10-bit external recording list, then you would have to add the Sony NX5U, EX1, EX3, and the new AF100A.
 
i am tired of statements like: best value for money, or cheap..... i did not buy it for that reason. i bought it because pany gave a "new" tech that no one else provided yet, but now its a past ...

for me when it comes to camcorder its not that much about the price, i cant say i dont care about price, i cant buy epic for example even at a reduced price (as i just cant afford all accessories), but give me good tech and a reasonable price and i will follow! if not ill have to give my hard earned money to sony or canon
 
The bottom line about the AF100 is that despite the price point, despite the great feature set and despite that on paper it should deliver, in reality it just doesn't deliver. For whatever reason it just doesn't!
Case in point: Shot with the AF100 and GH2. The shot included people indoors standing in front of a window during daytime. Exposing the faces properly with the AF100 resulted in the outdoors blown out. Not fully white mind you. Just the already known AF100 ugly blown outs. Exposing the faces correctly with the GH2 in the very same shot, shot at the very same time produced a much more pleasing outdoor. Isn't the AF100 supposed to have better dynamic range than the GH2?
 
You get what you pay for at the end of the day. And everything has its place.

I just lost out on a gig this week because the cost of the F3 kit I bring with me (even priced very competitively within my quote) priced me out of the client's budget. If I'd had a cheaper option to offer - even if it were a less capable option like the AF100 - I'd probably have landed the job, which at the end of the day, is far more important to me than being able to deliver superior image quality thanks to a better camera.

So all of these things have their place, you just need to figure out what matters to you.

For my part, I'm very keen for my BMC to finally arrive for this very reason - so that I can quote significantly cheaper equipment costs for jobs that can't afford the more expensive kit. It can open up new work opportunities for you - and that's the main reason to invest in any gear IMO.
 
Grug: There's truth in what you say, but I don't think we're talking about boosting the AF to $13,000+.. If they would greatly improve the camera and offer it for $7,500 I think a lot of people would jump at it because Panasonic still has a loyal following - I think... I'm shooing often with the HPX250 now because I can offer broadcasters 10bit 4.2.2. I paid $5,250 for that camera (it's gone up a little since). True I have given up shallow DOF, but it's an easier sell, and in fact most viewers don't really give a s__t anyway.

As I've said many times, I like the AF because it starts out very light and the MFT sensor means I can use light, manageable zooms... I realize that if they bump it up to take advantage of P2, like the 250, it will gain weight (and more grousing from those who think P2's outrageously expensive*), but it would have much more potential.

*And, BTW, I bought my five 16GB P2 cards in 2007 when they were nearly a grand apiece, and I've never regretted it because I still use them every single day... Just to put that in perspective, in 1996 I made a feature documentary in super16 and my film purchase cost (not including processing and work-print) was over $10.000! You can't re-use film.
 
Last edited:
You must log in or register to reply here.
Back
Top