AC130: Video footage of the Auto Focus in action

Hi Walter

I understood the area fuction (as long as you set it to iris/focus as an example) will sorta ignore all areas outside the square for both focus and exposure. By having that facility the camera has a smaller area to make both exposure and focus decisions so in my books it should lock in a lot easier and faster as it's only using around 25% of the image to make it's calculations. The reason I suggested a 50% zoom ( set the display to around 50) is that most people find that a 12X zoom is usually enough (unless you do wildlife or surfing videos) so if the lock in is greatly improved at half zoom you then have quite a useable range and you can then always switch to manual for those full zoom shots. Maybe do dual shots at say 25% and 50% ?? At weddings I usually use certainly not more than 6X or so during the vows and most of the other handheld shots are full wide and I must admit I never had lockups at full wide!! I did a similar test here on my 82's at 6X and 12X (it doesn't lock up of course and focus is instant) but practically do you really use the 160's full zoom range..I would also think that at a high zoom you have to be on a tripod anyway so then manual focus is a natural choice. For your choral groups I cannot see you going full zoom at any stage and probably a lot at 25% of the 160's zoom or less??

Rod will certainly appreciate your efforts!!

Chris
 
Here are test results of AF with full zoom, open iris and AREA On.

I agree with Chris this combination of settings is not something most videographers would find practical.

It can serve as a basis for a repeat test at 1/3 zoom.

Aside from the AF locking-up at infinity which was observed toward the end of the recording, having AREA ON, with no attention being paid to positioning and re-positioning of the area frame to high contrast areas in both scenes, is not representative of good use of the Area function.

So the lesson here appears to be don't use AREA On unless you position the frame on a high contrast object in your scene.

http://youtu.be/Za-zwyUWT2k
 
Last edited:
Wow that is a lot worse!! I really would have expected better results with the area fuction on. Surely the designers would test operations of a camera and lens before approving it for mass production?? Now, something I found on the 130 that I saw in the last clip was the AF behaves worst when it has to focus at a standstill. If you jiggle it and keep it moving it behaves better..I found that I was getting minor focus "misses" in a Church when shooting cutaways and a quite small zoom ...like 2X or 3X BUT if I kept the camera moving from one shot to the next and just deleted the in-between bit, the focus seemed to be better. For some reason it never used to like to focus on any stationary object for too long ...sorta "keep the camera moving" and it was a lot better.

The big issue is that you cannot expect your subject at an event to always be high contrast...we often have to film what is thrown at us, and it's far from ideal.

Well it certainly seems to me that your camera behaves much like Rod's camera which eliminates the faulty camera scenario. At least you guys know what to avoid now, thanks to all your hard work and tests Walter!!

Chris
 
kwkeirstead said:
Here are test results of AF with full zoom, open iris and AREA On....
http://youtu.be/Za-zwyUWT2k
Click to expand...
This is exactly what I experience but outside with plenty of light and contrast. Did you try this exact same test with a $200 consumer camera? would be funny if it would keep better focus.

But remember Jan (Panasonic Rep) said-- This camera cannot defy the law of physics with all it's high pixel count and it's large lens, etc... Even though the area focus box uses a small section of the chip (perhaps a 1/24 of an inch) and the internal moving lens travels no more than a 1/4 inch back n' forth. If it takes 30 seconds it takes 30 seconds. (based on timing your video starting at 18 seconds 'in', to the time of 48 seconds 'in' that the focus finally 'locked' in)

Walter we all wish Jan's Panasonic video, she linked to, actually addressed why the camera can not auto focus when there is plenty of light and contrast. Although they do say 'Professionals, like Bernie Mitchell, only use manual focus' and infer you're not a professional if you use Auto. Hey Bernie - people buy this camera specifically for auto focus and compact size. If this camera had a good depth of field I would keep it in manual but this camera's iris setting goes from wide open to nearly closed from one point of the horse paddock to the other which makes setting a hyper focal length not feasible to achieve good results. And we all know the law of physics of depth of field and f/stops settings. Add in Barry's own admission that the lens has poor sharpness wide open or nearly shut. I tried the infinity setting but when the horse gets about 30ft or closer I can see it goes out of focus. Let's be honest: all of Bernie's video samples --- every single videographer would shoot those scenes in manual and don't fairly represent the issue this camera has.

Personally I think this is not physics issue but rather poor design implementation. Ask yourself; when every single square inch of your image is out of focus for more than 20 seconds (with acceptable contract and lighting) what exactly is Piezo electric auto focus doing??????? for those 20 plus seconds???????? My answer: NOTHING!!!!!!! You can't tell me the physics of electricity (feel free to look up the speed of electricity) take 20+ seconds.
The shear concept of auto focus is: to be faster and more consistent than the professional user in manual mode while shooting a scene of ever changing focal distance in which the user does not have the time to "get focus" while doing other tasks. And this, my friend, is the law of common sense. If I can be adjusting the iris and then look out from my viewfinder & approximate the distance of the subject from the lens and then get back to my viewfinder and adjust my focus setting to that distance BEFORE auto focus does this then this camera is defective--- this is why this camera should be fixed. Auto Focus should be able to get focus before I could ever get my hand off the iris ring and lay it on the focus ring - BAR NONE!

I have a few more clips I will post later (I still have hours of footage to sift through and edit)
 
Here is a test done at Z50, open shutter, slow pan, 1080/60p, with AF on, and AREA on.

http://youtu.be/F71sLd3IIJQ

Barry's adaptation of Am Cinematographer's Manual says for 90 degree pan and Z50, if I am reading this right, (see page31 of Barry's book) the pan time should not be less than 26 seconds for 30p so, at 60p, who knows what it should be, but a rough guess might be 20 sec.

Notice that the camera seems to behave better when we pan slowly. The lens obviously has time to follow the panning when the pan speed is low. The result is less of a need to "jiggle" at the near/far targets.

Recording at 1080/60p probably results in a faster focus lock-in time compared to the previous recordings at 720/30p.

I don't think this brings us any closer to solving the "horse" recording issues but I am starting to learn how to use the focusing options on this camera as a result of various tests done.
 
Hi all,
Karl, thankyou very much for your tests. They seem to confirm that at least our two cameras cannot reliably focus on auto at near/ far scenarios.
My test was using two targets with reasonable range of contrast and using the function square. While not laborary scientific I kept everything as constant as possible. see http://youtu.be/nilQT3B3xtQ again if you have time.

This is a simple test which any camera sporting auto focus should be able to achieve. The fact that it failed on 30% of cycles, indicates to me that you can forget all the lens dynamics, pixel count , weight of lens etc. These may account for the lens being slow and I would accept that, but it works for 70% of the cycles, why not every time on an unchanging target?

What is becoming clear is that my camera doesn't meet specification because it cannot perform consistently- that is a cold hard fact. Here is a reverse analogy: I drive an automic car and while I rarely use the selector in manual, if it only worked 70% of the time, arguments along the lines of "well you don't tow anything so keep it in auto", or "it is a super auto box though so work round it" would not be acceptable by anyone- when I do want to use it I expect it to work if it is specified to do so..

I also object to the implication we hear that we are not professionals if we use auto. I am, and I occasionally do , and I need it to work reliably. This to me is just another "get over it" type of statement. My job or my hair colour has nothing to do with the way this camera works.

This is a fabulous piece of equipment and there is nothing this well equpped that comes close to it in the the price range. (I have looked!) However, it must meet specification, (something close to your heart Karl) and I don't see anywhere in the specifications, advertising or the manuals that says this camera should not be used in auto on max zoom. If it had , I could then have made an informed choice at least on whether, or not to purchase.

So thanks again Karl, Chris, ELN614, and all others who are contributing to this thread. Let's continue to be factual and objective, and not be swayed by opinion, or smoke screens of distraction. Our excellent NZ Panasonic have been immediately responsive in sending the video link to Japan for comment. I await feedback from them.
Rod
 
I thank you both for doing all the testing you have to show the flakiness of this camera.
Now that I see that the AC160A is as bad as my AC130. I will not be upgrading to the "A" version. I was thinking the update fixed this issue, but apparently not. We'll see how the new JVC camera works out shortly. I will get one next year after I see other people's posts on it.
Karl, I understand (see your quote)
kwkeirstead said:
Barry's adaptation of Am Cinematographer's Manual says for 90 degree pan and Z50, if I am reading this right, (see page31 of Barry's book) the pan time should not be less than 26 seconds for 30p so, at 60p, who knows what it should be, but a rough guess might be 20 sec.
Click to expand...
but unfortunately I don't have the luxury to set the camera to these setting and still able to shoot the horses.
 
Hi ELN614

Just for interest I actually posted a video of my HMC80's doing much the same test at both 6X zoom and 12X zoom http://youtu.be/P5UcIQwxEww If you haven't seen it the focus at 12X zoom from one object to another is instant and accurate and repeatable..in fact unlike the AC-130/160 it also stays in focus in between the pans too....I think I gave both my 130's a fair test ..I shot 80 hours on them over 3 months of weddings and Realty shoots and funnily the autofocus was only part of my problem.

I'm not telling you what to do or suggesting that Panasonic won't rectify this issue but I really see very little likelyhood of them doing so. My decision was to get rid of both my AC-130's while they were still relatively new ...I paid close to $4K each for them and I lost over $1K each on the resale....However I went back and bought another pair of the magic HMC80 and the loss from the 130's was a little softened cos the 80's are a lot cheaper. Hopefully you will get some results from Panasonic but if you have no success with them don't beat yourself up over it and move on.

Chris
 
Chris,
I actually own an 80 and is part of my deposition setup #2
A 12x lens is worthless in a large horse paddock.
I need 16x very minimum. 18x is better. 20x -22x is ideal.

thanks anyways.
 
No problems..I wasn't suggesting using an 80, just comparing the fact that it works as it should in all respects.

Just for interest do you still have the JVC 700 ??? That would work pretty well if it had the Fuji 17X lens on it but it probably has the Canon 14X ???? Have you tried the exact same scenario in the paddock with the JVC camera??? I was just wondering if that might also lock up at full zoom??? It might add a little more evidence that the 160 does not focus at 22X!!!


Has anyone else found camera with a BIG zoom that autofocuses correctly and quickly at 22X??? Can it physically happen or is the magnification just too big for autofocus to work consistently??

Chris
 
In cameras available now, the Canon XHA1 is 20X, and works very well indeed. That is actually an extremely good camera, the lens is great, focuses fast, and is a 20x. It's availble now, and at decent prices. Next to that is the Canon XF300, which is an 18X. That's all I know of...
 
Last edited:
I'm posting this to complain about the notion that the focusing problem is due to "the laws of physics" given the zoom range and high pixel count.

Bull! ... I have a PhD in experimental Physics, and worked on a Nobel Prize winning research discovery. The zoom range cannot be an issue given adequate servo control, and the high pixel count only adds to the information necessary to achieve focus.

The problem has to be in the software architecture or CPU processing power. I use an HPX 250 in autofocus mode and find the autofocus similar to (or better than) an HMC150 or XH A1. As I understand, the HPX250 shares the same lens and imager, but different signal processing and design groups (hence different CPU hardware and software designers). Theoretically, software limitations can be fixed, but at this stage I doubt it would happen. Hopefully new cameras will autofocus properly, since I'm sure Panasonic is aware of this shortcoming.
 
Great! now we have a scientist on the team! Welcome to the auto focus discussion Terry, your input will be gratefully received by the active members on this thread and possibly others who would like another view (other than ours) . Good to hear from you!
Rod
 
NEW TEST RESULTS - AG-AC160A, on auto, Z50, open lens, 1080/60p, AF ON, AREA OFF, near object at 4.5 ft, far object at 37 feet


CONCLUSION: In this highly structured test, not representative of field recording, my AG-AC160A was NOT able to consistently achieve a focus on the near object. If you nudge the lens (very small pan away from / back to your near object), the camera will almost always lock-in to the near object. Once in a great while, no amount of nudging results in lock-in and the remedy for this is go back to the far object where you will almost always get a focus and then return back to the near object.

COMMENTARY:
The test will be repeated with AREA ON, going NOT in a straight arc from a variety of near/far objects.

The expectation is that focus lock in performance will be significantly improved.
 
Last edited:
ChrisHarding said:
...Just for interest do you still have the JVC 700 ??? That would work pretty well if it had the Fuji 17X lens on it but it probably has the Canon 14X ???? Have you tried the exact same scenario in the paddock with the JVC camera??? I was just wondering if that might also lock up at full zoom??? It might add a little more evidence that the 160 does not focus at 22X!!! ...
Chris
Click to expand...
Chris it is an ENG style camera. I have the 17x lens which would work. FYI he lens is all manual and would work fine. The issue is to invest in 1200+ dollars worth of batteries/chargers. I have used ENG style cameras (Panasonic supercam 800, and DVCPro 410a) but I was getting very tired of moving a large camera around and those cameras are now in semi-retirement. (used 4 days out of the year)
 
Terry_Martin said:
I'm posting this to complain about the notion that the focusing problem is due to "the laws of physics" given the zoom range and high pixel count.

Bull! ... I have a PhD in experimental Physics, and worked on a Nobel Prize winning research discovery. The zoom range cannot be an issue given adequate servo control, and the high pixel count only adds to the information necessary to achieve focus.

The problem has to be in the software architecture or CPU processing power. I use an HPX 250 in autofocus mode and find the autofocus similar to (or better than) an HMC150 or XH A1. As I understand, the HPX250 shares the same lens and imager, but different signal processing and design groups (hence different CPU hardware and software designers). Theoretically, software limitations can be fixed, but at this stage I doubt it would happen. Hopefully new cameras will autofocus properly, since I'm sure Panasonic is aware of this shortcoming.
Click to expand...

Thanks I sorta knew this but never came out and said it assuming the board Mods would delete my post and ban me. My dad is a retired Optical Physicist, well published and would work on top secret gov't applications. (he would hint on how there are better quality spy images that are not being released.. not that he knew about them, but rather the capability of hardware which was made) I knew enough from him (that being when growing up around optics, etc..) that Panasonic was not being completely forth coming... that is why in my post I said it is technically impossible and went on to explain why there was a discrepancy. But, as expected, it was blown off.
When I go home in January we will talk about this camera and I will get his expert opinion.
I think the key problem is under performing Piezo system communicating with the servo motor control. But good luck getting them to fix that. I suspect a firmware update of Piezo would be the solution since we know the servo can move fast when hitting the auto button.


Side note to growing up with an optical physicist: I thought it was normal that all kids built their own telescope and ground their own lenses. nothing like hours of hand "grinding" with colored powder which would make a mess. We also had microfiche machines everywhere and he assembled them in the house.... mom was pretty perturbed about that business. My dad also probably came up with the first 3D video camera and recording using one lens consumer level recording equipment. Several years back when I went to NAB and they were pushing 3D recording... my thoughts where "took you long enough". He made his first prototype somewhere around 1985-87. It was a crude but viable. He also had a hand into one of the very first tv based game. Back in the late 70s or early 80s (pre Colecovision) ... do you remember the game where you would point the gun at the tv which had cube bouncing on the screen and you would point & shoot and get point each time you would shoot the square --- If I recall correctly he help the company fix an optical problem within the gun.
 
If anyone is interested, I'll offer a speculation of what is behind the AC160/130 focus issue.

Please be advised that I have no official information... just pure guesswork on my part. I do have great respect for Panasonic video gear and for Jan who graces this forum with real info. Also, as a Panasonic shooter, I have a golden statue of Barry Green in my studio which I worship daily.

As I said before, I have 19 years experience as a Physicist in the academic world (got to play with anti-matter and live to tell about it), and later I was a R&D director for a large corporation.

Here is what I think happed, primarily based on my corporate experience. Given Panasonic's proven P2 pro camera products, and the emerging DSLR factor redefined in the AF100, it would seem a good idea to build new cameras which would develop a new common lens/ imager combination and then leverage the existing back end of both the P2 line and the AF100. I'll assume there were separate development groups for P2 and AF100 gear. Quite frankly, this seems like a great idea to minimize R&D costs, yet deliver a wide array of new gear.

Assuming some time pressure for R&D, the quickest path to a new product is to incorporate as much existing hardware from the P2/ AF100 as possible.... and that brings us to the CPU chip that processes auto focus software. (I'll assume it is in the image processing chip, but don't know for sure).

Now let's think of how these little miracle chips have improved over the last few years. Consider a modern smart phone compared to last year's model... faster, better, more pixels, mainly because of rapid CPU hardware improvements.

I'm betting... (maybe a Canon HV20), that a CPU chip inherited from the AF100 was not fast enough to handle the increased auto focus workload. I can just imagine an engineer dealing with this problem begging for a newer chip, but facing management unwilling to accept delays or cost overruns. They would say, "we have to make do with what we have now, and after all, real men use manual focus".

Bottom line, I believe the signal processing hardware is about one generation too old to handle the workload demanded by this lens/ imager, and maybe even the software was tweaked to deal with the lack of horsepower, and ended up with some quirks trying to hold focus as we see.

I have followed this thread from page 1, and although normal circumstances can easily screw up auto focus, there have been examples posted here that clearly indicate it is unreliable in repeatable circumstances. Even though there is a smaller auto focus thread in the HPX250 section, I remain pleased with HPX250 auto focus. Assuming the 250 has a different processing chip and software inherited from the P2 line, it seems at least up to an average industry performer.

If it were only a software coding or algorithm error, it theoretically would be fixable, but in the real world it is unlikely that any company would spend the money to track it down.

To end on a positive note, the real fix will be in the new generation of processing chips, which are certainly in the newer cameras. I'm looking forward to getting an AC90, and I'll bet that old HV20 that it won't have similar auto focus limitations.
 
Hi Terry,
If anyone is interested? you gotta be kidding! That is most interesting. How many 130's 160's and 160A's sold to date? We are all very interested.

As you will have read I have forwarded my findings to Panasonic and await their feedback. I, along with many others also have great regard for Barry and Jan, but assuming Panasonic read this thread it is pretty obvious that there is some dissatisfaction with this otherwise superb camera's ability to perform a simple cycle consistently and shouldn't be defended on the basis of" it is what it is". It is not performing to specification and should be addressed by the manufacturer.

I have every confidence that an organisation the size of Panasonic take these issues most seriously and I look forward to some positive reaction from them.
 
Hi Rod,

Thanks for your interest. As one of my politicians once said "I feel your pain".

I'll lend my voice to say:

1) Based on the evidence in this thread, the 130/160 cameras do NOT autofocus with the precision and reliability expected of even the previous generation of cameras. I would consider it non functional for professional recording requiring autofocus. (And yes, it is a wonderful camera otherwise).

2) Based on my experience, the HPX250 does achieve a reasonable standard of autofocus quality relative to current industry standards. Yes, it hunts more than I like in low contrast, but I have to say it has a useful autofocus.

I'd be willing to argue these conclusions way into the weeds of pixels, photons, or software code.

And one more thing. Quality autofocus in video is not an easy job. It's no big deal to put a servo on a lens, and contrast based focus detection algorithms are probably standard. The key is how efficiently the image data is processed in the cpu hardware since it needs to figure out what to focus on within all the frame data. If the hardware is marginal and dosen't keep up with the data flow, well then control commands are not in time and then....well you know.

This has to be a serious problem if my hardware limit theory is real, since you can't just change a cpu. If I were in charge, I would try mitigate the problem with an unusual software approach. I would establish two distance setpoints to limit the allowable focus range. Constraining the focus range could help streamline the calculations needed in the CPU. The XH A1 is generally considered good at autofocus, but is HD and has a really old CPU. The difference is that it has a separate IR rangefinder that simplifies the focus work. (Note, try shooting an XH A1 through a glass window and it won't focus on anything.) Focus range setpoints are not ideal for everything, but would enable more professional applications.
 
Terry, (if you know) what would explain why this camera can shoot a stationary scene 120+ feet away ( camera lock down) and while on auto focus it will consistently set it's focal distance of a few feet away (2-5ft - I think everyone saw my earlier post/image of the barn.) With no objects in the framing to be closer than (lets say) 35ft. ... I should add the camera is zoomed out wide.
 
You must log in or register to reply here.
Back
Top