AC130: Video footage of the Auto Focus in action

[Mike Warren]

I've read the whole thread, and many others, both here and elsewhere. I lurk a lot, but don't post often.

Based on your videos, I'm convinced that your camera is behaving the same as the other 130/160s.

Here is what the camera sees as you pan over the orange:

Sometimes, it will see enough contrast, particularly when you get a bit of the edge of the orange in the area and attempt to focus, but other times it will decide it doesn't know what to do. If I was filming that, firstly, I would not expect to cover both subjects in the same shot unless I was manually pulling the focus. There is too much risk of hunting or overshoot with any auto focus system I've ever used. So I'd expect I would be editing out the section in between. Therefore, I'd first place the focus area over something like the edge of the bowl, but not the horizontal edge. I suspect that these cameras will not look at changes in the vertical axis as that would take more processing power. It would only take me a few seconds to prove that. Then I would move it on to the proper subject. That's if I really had to use it in continuous auto focus. Ideally, I would want to use the push focus function because it's rare that my main subject is in the middle of the frame.

When you hit the curtain in this position it focuses more quickly:

When you hit the curtain in this position it focuses more slowly:

That makes sense because while out of focus on this shot, there is not much contrast:

When you shake the camera you are just changing enough of the data to trip the algorithm into action. You wouldn't need to bump the camera. Merely reframing slightly lower would probably do the same thing.

All IMHO, of course. I wish I had access to one of these to test it myself.

 

[ELN614]

But I am curious about one thing. What was happening just prior to the start of this clip? You are fairly wide and yet quite out of focus.

I moved the tripod to a different place in the field. at the time I was busy leveling the camera head (takes about 10-15 seconds to get the bubble perfectly in the middle.) After I level it I look up and see this terrible out of focus image and hit record. (But this is not always the case.)

As you notice both clips start on something it could easily focus on while I got the camera leveled but isn't.
You need to know this happens only 1 out of maybe 15x when I do this. So the other 14 there are no problems. Plus you need to know it will lose focus in the middle of shooting the horse as it's moving. Plus you saw the first video (on page 1) of the pole that covered 75% of the frame and the camera never noticed it until it was 100%.

If I zoom in and slow down my 'good' footage I can frequently see the focus being a little soft. That is not my issue. I understand auto focus is not perfect. My issue is when the image has everything far far far out of any type of focus.

 

[ChrisHarding]

Hi Guys

I do agree with Barry that it's a different camera and often your expectations will be disappointed cos your "old camera never did that" ..Coming from an HPX 250 I suspect would indeed be a lot easier as they are similar animals. I admit I was thrown by the manual handling of 4 ND filters cos my HMC's had auto ND ...it was a pain but I managed. Now if my previous camera had been an HPX the only change would have been the recording format ...I guess it's a simple matter of choosing the right camera for the job!!

One suggestion though would close this thread and make most people understand would be a test video from a kind HPX 250 user that would replicate Rod's couch/orange focus test.... and maybe a long zoom shot of moving objects to replicate ELN614's horses issue. Surely if current 160 owners could see that, yes, the HPX 250 also does the same thing at full zoom then they at least know that "that's the way these series of cameras work" and they can adapt their shooting skills around it.


Anyone here got an HPX 250 that is prepared to replicate the test videos ????


Chris

 

[Mike Warren]

I moved the tripod to a different place in the field. at the time I was busy leveling the camera head (takes about 10-15 seconds to get the bubble perfectly in the middle.) After I level it I look up and see this terrible out of focus image and hit record. (But this is not always the case.)

Something must have made the camera focus really close. Perhaps you were in front of it while leveling it.

The loss of focus while zooming in to the horse running away from you is more of a problem. My guess here is that the lens on the cameras is not parafocal and the software has to compensate dynamically as you zoom. I've see that in other clips from these cameras and the 250. In fact, I noticed it in one of the first HPX250 videos I saw posted online.

You need to know this happens only 1 out of maybe 15x when I do this. So the other 14 there are no problems.

Of course, it's difficult for me to do any more than speculate without hands-on, but I've not seen anything in posted clips that I can't see an explanation for. I realise that doesn't help and should probably have just remained quiet as I have been doing since the beginning of all these complaints.

Plus you saw the first video (on page 1) of the pole that covered 75% of the frame and the camera never noticed it until it was 100%.

I actually thought it was quite good to do that. There was almost no contrast in the pole while out of focus and the camera decided that the already focused subject had priority. Alternatively, what it could have decided was to attempt to focus on the pole (assuming the pole ended up being completely featureless; remember, the camera doesn't know what the object is or if it is in front or behind the current focus plane). If it turned out that what we knew to be a pole was actually a wall behind the horse, the camera would have hunted. It's a compromise. The processors used in things like camcorders are not very powerful. There is limited code space and speed, especially since everything has to be calculated in the time-slice of the fastest frame rate the camera is capable of, but can only act on that data at the speed of the currently set frame rate.

 

[kwkeirstead]

Mike, on the contrary,

"..Of course, it's difficult for me to do any more than speculate without hands-on, but I've not seen anything in posted clips that I can't see an explanation for. I realise that doesn't help and should probably have just remained quiet as I have been doing since the beginning of all these complaints."

I feel your intervention and commentary has been extremely valuable.

I am now sensitive to looking for high contrast areas on the approach to a subject.

I did a test back in July involving a near object (anchor) and far object (church steeple) and whereas I did not understand the results at that time, your explanations account for the results.

I predict as the 130/160 owners learn how to make best use of these cameras, they will have few complaints.

 

[ajcourtney]

Man, what's so hard with grabbing the focus ring and pulling it yourself? Coming from using DSLRs exclusively for the past three years, the 160 sounds like an absolute dream of an event cam! Take your sdi outs into a BMD ATEM switcher, and you're all set for live HD switching. For all the complaints about these cameras, I'd suggest you run and gun solely using DSLRs (at apertures of 2.8 and higher rogar-Shock(DBG) for professional work for a solid 12 months and then revisit these issues, LOL. I think your perspective will change massively.

 

[ChrisHarding]

Hi AJ

It's not the fact that these guys cannot use autofocus, I'm pretty sure they are most capable of shooting everything in manual BUT their issue is that since the camera offers AF then by rights, it should work...like buying a car with auto transmission and having to shift manually cos it doesn't change gear sometimes. During the 3 months I had 130's I was told "you should be shooting in manual" too!! It's more of a principle thing now ..."It has an auto button so it should work"

My exact reason for not using DSLR's at weddings....manual zoom and manual focus and trying to follow focus as the bride walks down the aisle...I have great admiration for you guys..I couldn't do it!!!

Chris

 

[Nate Haustein]

Even after 19 pages, I agree with this.

Man, what's so hard with grabbing the focus ring and pulling it yourself? Coming from using DSLRs exclusively for the past three years, the 160 sounds like an absolute dream of an event cam! Take your sdi outs into a BMD ATEM switcher, and you're all set for live HD switching. For all the complaints about these cameras, I'd suggest you run and gun solely using DSLRs (at apertures of 2.8 and higher rogar-Shock(DBG) for professional work for a solid 12 months and then revisit these issues, LOL. I think your perspective will change massively.

For most cases, MF is simply the way to go, if your camera has autofocus or not. Not to mention the DOF of the AC160 is deep enough for quite a bit of wiggle room. No matter what anyone says to the contrary, we all know by now that the autofocus on the AC130/160 can be pretty crummy, it just depends on if you think the other "Pros" to the camera outweigh the "Cons." For me, they do, but you can bet that I keep that switch set to manual when I'm out doing my thing. I understand you guys are looking for the reason behind things in a scientific sense, but I don't think that there's some magic process to making the AF work. In my opinion, if you can't simply point in full auto and achieve focus, there's not much point in trusting it.

 

[Terry_Martin]

Barry,

Clearly my comments have gained more attention than I thought.

You are the last person I would want to argue with in a video forum, and we are in substantial agreement except for some physics nit picking. Of course you and Jan are correct that the lens is more massive, and the sensor resolves to finer dimensions. I support all you argue, except that it is an engineering issue, not a physics issue. If you use a HMC150 servo on an AC160 lens, then there could be problems... but that's why you get and engineer like Mike to design a capable AC160 servo. In 1987 I was tasked with designing a piezoelectric motion control system for a quadrupole magnet used to focus an electron beam. The magnet weighed about 1/2 ton, and the position tolerance was about 1/100 the size of an AC160 pixel. I won't mention that it cost more than a few hundred cameras.

Our only difference is that I believe focus performance is actually limited by engineering and economics rather than physics. If physics were the limit, then no camera in this class could ever focus better. I bet in a few years you will be writing about a Panasonic camera that focuses flawlessly in the dark, it just won't focus faster than the speed of light.

I first read about auto focusing at 24p being harder than 60i from something you wrote, along with a lot of other video gems of wisdom. This is a most legitimate attribution to Physics, since reducing a sample rate always makes a measurement more imprecise.

Another point of agreement: I stated that I am fine with the HPX250 autofocus, but on one occasion I had to shoot a ballet recital in a rehearsal hall. The floor was grey, the backdrop was light grey, some dancers had drab costumes, and of course the lighting was horrible. The HPX250 autofocus seemed to hunt a lot. I just switched to manual focus because I could see that it would not be fair for any camera to focus on a set of dancing grey cards.

I only jumped in here to try to be helpful, and because there seemed to be a lot of focus examples that seemed foreign to my HPX250 experience. Also, every one should understand that my comments linking the focus problem to an underpowered CPU is just my opinion, and it is based on a number of technical assumptions that are unverified. Also, I'm glad Mike Warren jumped in with his engineering experience.

I'll be traveling for the next week, so I will go dark after this post for a week. When I get back, I'll try to duplicate the orange/ curtain shot with a HPX250, although lighting and other differences will make this a test for curiosity, and not too scientific.

 

[ELN614]

Something must have made the camera focus really close. Perhaps you were in front of it while leveling it.

That's a big no. the bubble is the back and I hold the camera's handle when leveling. Lens is always pointing away from me and camera is always parallel to the horizon. (ie it is never pointing to the ground or pointing to the sky.)

 

[ELN614]

Man, what's so hard with grabbing the focus ring and pulling it yourself? Coming from using DSLRs exclusively for the past three years, the 160 sounds like an absolute dream of an event cam! Take your sdi outs into a BMD ATEM switcher, and you're all set for live HD switching. For all the complaints about these cameras, I'd suggest you run and gun solely using DSLRs (at apertures of 2.8 and higher rogar-Shock(DBG) for professional work for a solid 12 months and then revisit these issues, LOL. I think your perspective will change massively.

Because you can't adjust the focus ring and iris ring at the same time.(and get solid results) This camera can change from open to 5.6 from one end to the other.

I can assure you no professional in my specific industry would be stupid enough to use a DSLR in a horse paddock (for video).
I would love to see you come out with your DLSR and try to keep the same framing of the horse from one side of the paddock to the other as it passes by you. (ie zooming in and out constantly) Paddock is around 200 ft long. all while you try to keep your focus and maintain your correct aperture. Remember one hand will always need to be on the handle for tilting and panning. I'll even let you use 6.0 or higher on your DSLR.

 

[ChrisHarding]

Right on!!

We have lots of instances where we just have to have some auto functions to get the shots we need because we just don't have enough hands to do it in full manual. When we need auto functions that are present and available on the camera we do expect them to work. When I approached this issue way back in April this year I was told the same thing..."use manual" ... if I had wanted a manual camera then I would have bought one ..but I wanted a camera that had both auto and manual and yes I use both all the time..I even use ALC on my audio at wedding receptions during dancing cos the audio levels change so often... we need both facilities at certain times to do our job correctly so when a function doesn't perform as intended we like to find out how to get it to work correctly by getting advice on the forum, not how to bypass it. We know how to do that already! If your washing machine stops working do you wash your clothes in a tub in the back yard? of course not, you call the repairman and get the machine fixed...that's what is trying to be attempted here .. to rectify an autofocus issue on the camera.

Chris

 

[ajcourtney]

Thank you,ELN614, that is exactly my point! Regarding iris pulls, IMO there is no need for that unless you are transitioning from massively changing lighting conditions, like from interior to exterior. I have no problem seeing temporary reductions in ambient lighting conditions due to passing cloud cover. That is exactly what happens with our own eyes, so it's not an unnatural occurrence. What i find absolutely maddening is fluctuations in background exposure due to the autoexposure being fooled by changes in framing composition. And the only way I know how to prevent that is to lock off your exposure with a grey card, which is pretty much what I do for every shot, even run and gun. If none of this is acceptable to you, then grab a gh2 with the 20/1.7, a few NDs, put it on sticks, push record, walk away, and now you've got yourself a nice wide to cut to while you work with your130/160.

 

[ajcourtney]

Chris, there is only so much one can expect from technology. You mention the audio level fluctuations at weddings. Short of having a dedicated sound recorder with someone riding those gain pots or faders, you're just not ever going to achieve perfect results. We have been conditioned by Hollywood that perfection is now average. Its not! There's a reason why there are 300 names listed in the credits, lol! IMO, it's asking a bit too much from any piece of gear to expectit to perform the roll of multiple humans.

So in your case, you're just going to need to set your levels using maximum expected peaks so you don't clip, maybe engage the limiters if they're ok, and just fix things in post using your ears. This is just the way we all have to work even with dedicated gear/people. If you're not mixing on an external control surface or mixer in post, I would strongly suggest you look into that as it will make the process much more enjoyable. Lastly, pick up a copy of ozone for audio mastering and that will help greatly in achieving a polished final product. You don't need to totally smash your stereo mix down track using the limiter either. Just knock a bit off the top and you'll preserve your dynamics yet still raise your overall track level by a few db's to makeup for the natural level fluctuations at your event (assuming you've already ridden the faders during mix down)

Https://www.youtube.com/watch?v=VobieYuEFKM

This is a recent mostly pro bono piece I produced that illustrates the results you can achieve when you follow what i wrote above. It's not perfect but it's perfectly acceptable. This guy's dynamic range was probably a solid +-18db's (total 36db range) that morning. So I knew that and just had to set it to peak without clipping and try to make up the difference in post. IMO, I achieved solid results. There is some noise creeping in during the soft passages because I'm making up close to 20 db's in some sections in post and due to a crummy wireless mic.

 

[ELN614]

... What i find absolutely maddening is fluctuations in background exposure due to the autoexposure being fooled by changes in framing composition. And the only way I know how to prevent that is to lock off your exposure with a grey card, which is pretty much what I do for every shot, even run and gun. If none of this is acceptable to you, then grab a gh2 with the 20/1.7, a few NDs, put it on sticks, push record, walk away, and now you've got yourself a nice wide to cut to while you work with your130/160.

Several things
1) I gave the previous example of Open-5.6 as very common thing to happen. Add in the fact a cloud may brush by and may other variables. There are some days I am constantly flipping the ND filter back n forth. so "locking off" exposure is not going to work.
2) The horse runs in a paddock 100x200 ft (approximately) Locking down on what??? where??? A horse does not know how to stay in the frame. Add in the fact is you need very specific angles. (close-up side shots and close-up straight on front & backs) Any lock down is TOTALLY worthless. Plus if I was to have a bunch of 'cut to' clips that were wide shots, my client would never call me back. I am sure your next suggestion would to get a whole camera crew. I could kiss profit margins away if I start bringing in a camera crew because an Auto Focus camera can not focus.

So you all understand: I get my shots!!! This auto focus does work the majority of the time. It is the few times which makes you pause and ask what is wrong with this camera. It has me reflecting back to the last 18 years of taping horses and saying "I have never ever had any of my Panasonic cameras go out of focus, like this, with no reasonable explanation". PERIOD.

There are many samples in other thread that show this camera going out of focus for no apparent reason. It is just this thread has shown outdoor images in good contrast still have the auto focus failing miserably.
Well I am just saying they better find a new approach focus if they can't do it right this way.
If Panasonic made a camera knowing the long large lens was can't auto focus like previous models... they should of made a shorter lens... perhaps 20x or 18x until technology was advanced enough to not fail.

 

[ajcourtney]

You expose for your primary subject, and let your highlights and shadows fall where they may. If you don't your horses are going to be changing colors/shades based on their framing. And I guess we just have different opinions on multiple camera angles. Single angle shots are completely non-existent today on broadcast television.

I'm guessing the lens is not a constant aperture zoom?

 

[ajcourtney]

It's not a fixed aperture zoom - it's 1.6 to 3.2. If you were filming indoors, then i could see your point. But outdoors, the absence of constant aperture should be a moot point. I know in broad daylight with the GH2s @ 1/40th & ISO 160, I'm beyond f22 - maybe f22 @ 1/80th. So I'm stacking NDs (.9's or 1.2's) with a polarizer to get down to ~f8.

The 130/160 can't be much more than a couple stops slower than the GH2 I would think; so that would still leave you @ f4 with your ND engaged which makes the lens essentially a constant aperture zoom once locked off.

 

[ELN614]

Single angle shots are completely non-existent today on broadcast television.

This is not for broadcast and you have no idea what I do.

Please stop responding to this thread.

2 minutes of my life (responding back to your last post) I will never get back.

 

[ajcourtney]

You're right, I don't know your use of this camera. But as its been said before, if this cam doesn't work for your use, find one that does. I'm simply providing an alternative viewpoint for anyone who cruises by this forum and gets the impression that this camera is really screwed up.

Its pretty childish to disrespect 31 years of experience. I did take a look at your posted video and in certain areas, I am actually seriously impressed with the autofocus on your camera. That shot of the horse running away from you is mostly incredible. Short of using sonor, I don't know how the camera pulled it off. That is an absolute worst case scenario for pulling focus without constant racking. I suggest you try to repeat that performance on manual. Even after decades of manual focus experience, I would be seriously challenged by that shot. It looks like camera had trouble in the middle before deciding to just dump the lens to infinity.

 

[Rod Pike]

You're right, I don't know your use of this camera. But as its been said before, if this cam doesn't work for your use, find one that does. I'm simply providing an alternative viewpoint for anyone who cruises by this forum and gets the impression that this camera is really screwed up.

Its pretty childish to disrespect 31 years of experience. I did take a look at your posted video and in certain areas, I am actually seriously impressed with the autofocus on your camera. That shot of the horse running away from you is mostly incredible. Short of using sonor, I don't know how the camera pulled it off. That is an absolute worst case scenario for pulling focus without constant racking. I suggest you try to repeat that performance on manual. Even after decades of manual focus experience, I would be seriously challenged by that shot. It looks like camera had trouble in the middle before deciding to just dump the lens to infinity.

I have seen a lot more of ELN614,s work and Chris's and believe me it is indeed stunning. This thread was started to discuss an issue many of us are having with this camera. Like ELN614 and Chris, we are sole traders and have to work alone in a competitive environment. This thread is about the camera , if you wish to discuss how to shoot broadcast with a team and how to spend money on more equipment please start another thread.

The input and readership on this topic has been massive and everyone is on topic and giving proactive input. Please let's keep it that way. Thankyou.