[Vylen]

I'm also interested also to see what the max length of one clip is on this camera. The K5 of course was limited by the 4gb limit of the FAT32 file system of SDHC cards, but this one is designed to use SDXC from the start. The specs you have listed say "up to 25 minutes," but it is not clear if that is for any format, which would indicate the time limit is designed around the EU 30 minute import duty restriction (which raises the question, why cut it off 5 minutes before the limit?) or whether it is a function of max file size again.

Looking at the chart, it does appear that at 720p, 24fps, the 4gb capacity is well above 25 minutes, so I surmise it's probably the first scenario.

Could just be a sensor heat limit as well...

[Dave L]

PENTAX K-3The new flagship of the PENTAX K digital SLR camera series, featuring an array of new technologies including a new anti-aliasing simulator?RICOH IMAGING
Just scroll down to "Storage capacity/<Movie>"...
The 16 min 14s video is about 4GB, as you can see, but of course there is a sound recorded as well...

EDIT:
I just double checked the specs for the K-30/K-50 and they say 4GB is 16:21 (1080p30), so you're right it should be roughly the same.

Wow! But according to that, it's you who's right. So 33Mbps it is then, according to Ricoh - of which around 500k to 1.5Mbps will be audio, so about 32Mbps for the video at 1080p/30 or 1080i/60. Mighty impressive if true - though somewhat limiting in video length, at less than 17 minutes for 1080p/30 unless certain cards can be formatted in ExFAT, for example. I'm uncertain whether or not the Quicktime container also has a 4GB limit.

So unless the bitrate is user configerable, it looks like to record continuously for the maximum 25 minutes, we'll have to drop to 720p/30 or below. K-3 will record for a lot longer than K-5 though, at any given HD resolution.

[txsbluesguy]

You would be surprised how often even a broadcast Network Evening News segment is done with local talent and one universal camera/audio/light technician - and the "Producer" is the assistant.

Having been one of those camera guys that shot ENG style and local spots with a crew of me and a Reporter/Producer or me and a Producer/salesperson ..... I can honestly say, I really don't know a single one out of the dozens that speak to almost daily that would ever trust auto focus ..... There's too many ways it will screw up in ways that are too noticeable. In my conversations I have, they HATE the modern fly by wire lenses of the mid size camcorders and proper ENG shoulder cameras still don't have AF.

The size of the crew has little or nothing to do with focusing that's all on the camera guy . Proper technique and shot selection pretty much eliminate any need for AF.

[johnmflores]

You would be surprised how often even a broadcast Network Evening News segment is done with local talent and one universal camera/audio/light technician - and the "Producer" is the assistant.

Having been one of those camera guys that shot ENG style and local spots with a crew of me and a Reporter/Producer or me and a Producer/salesperson ..... I can honestly say, I really don't know a single one out of the dozens that speak to almost daily that would ever trust auto focus ..... There's too many ways it will screw up in ways that are too noticeable. In my conversations I have, they HATE the modern fly by wire lenses of the mid size camcorders and proper ENG shoulder cameras still don't have AF.

The size of the crew has little or nothing to do with focusing that's all on the camera guy . Proper technique and shot selection pretty much eliminate any need for AF.

Yup, I'm familiar with those crews. That's one of the advantages of the smaller chip cameras-more DoF and thus more margin for error. Experience and smart shot selection go a long way for sure. I often shoot my GH2 with MF Pentax primes and try to keep reasonable apertures to allow for some DoF and some safety margin.




The M28 F3.5 blew me away on this shot, it was such a sweet image.

and here's another one:




But I recently purchased the Panasonic 12-35 F2.8 and have been experimenting with the camera's face detect AF. It's quite good and I hear that the GH3 is better still. I do set up a second cam in these situations in case the face detect AF hunts, which also gives me b-roll for some of the longer interviews that I'm shooting. I'm looking at the K-3 and wondering if it can fit in my workflow.

It's funny how DSLR shooters are currently so enamored with razor thin DOF. I was surprised to see PBS Frontline make a rookie error-too thin a DOf for a subject that liked to move around a lot, What Blog is This?: Razor Thin Depth of Field is a Harsh Mistress

[repaap]

I don't really like too shallow Dof with videos. You have to be really good to make it. And in best cases It might ruin your shot, and you will be left with hoping for next opportunity. It is good with more time and many attemps to make a special effect like shallow Dof. But with APS-C you can use faster and little longer lens also for that, no?

[TaoMaas]

It's funny, after shooting some video myself, I totally appreciate the long list of credits at the end of a movie now!

Hey, you should do what we do. We have a decent list of credits for our show, but we just repeat the same 4 names for everything.

[TaoMaas]

I can honestly say, I really don't know a single one out of the dozens that speak to almost daily that would ever trust auto focus ..... There's too many ways it will screw up in ways that are too noticeable. In my conversations I have, they HATE the modern fly by wire lenses of the mid size camcorders and proper ENG shoulder cameras still don't have AF.

The size of the crew has little or nothing to do with focusing that's all on the camera guy . Proper technique and shot selection pretty much eliminate any need for AF.

I work with a guy like this on a travel show. His background is as a very traditional news shooter. I came from the world of commercial production. I tell our interns that the difference in our approaches is that, "he has ethics...I don't." Not literally, but we do go about things in different ways. I just mean that he has a fairly rigid set of rules he goes by, whereas I'll do whatever works. We have several cameras available for use. Our primary camera is a traditional ENG camera...manual focus, designed to be shot off the shoulder or sticks. Our other main camera is a high-end auto-focus camera. It used to be a Sony DCRVX-2000, but now we have a Panasonic AG-HVX200A, I believe. My co-worker absolutely refuses to take the little auto-focus camera out on shoots. He'd much rather work with the big camera. However, I use the little camera all the time. (He and his news buddies make fun of me for it, btw...in a good-natured way. lol) You're right about not trusting the auto-focus. It could definitely burn you if you needed to hold focus on something that wasn't centered, like your typical interview. With the little camera, I started doing something unknowingly that my co-worker just realistically can't do with the way he's chosen to work. How I found out that it was different was when one of our reporters took the little camera home one weekend so he could shoot some b-roll of a festival he was going to with his family. We were looking over what he'd shot and he said, "Look...I even threw in a TaoMass close-up!" What he'd done was simply zoom out wide and stick the camera 6-8 inches away from the subject. He was right. I do that all the time with the little camera. And while I know you can also do this with the big camera, it's the shooting situations that make it impractical. One of the places I do it most often is inside a restaurant kitchen. It's often very cramped quarters and we try to get there during rush hour so there's a lot of action. It's my job to document things while staying out of the way. I'll shoot my wides, mediums, and tights...but then I'll also do what that reporter was talking about...zoom the little camera out, stick it as close to the chef tossing roasted veggies over a flame as I can, and let the camera do the focusing. Typically, he's done in about 10-15 seconds so you don't have long. You know how a teenage girl with an I-phone can get a closer picture of the paint job she just had done on her fingernails than someone with a FF DSLR shooting a non-macro prime can? Yeah...same thing. I just see it as using the right tool for the job. Now...this same guy has no problem using a GoPro or shooting with his DSLR, so go figure. lol

[jaad75]

Wow! But according to that, it's you who's right.

In theory. But I don't believe those numbers can be converted to bitrates so straightly - just take a look at DNG files capacity:
there is 18 DNGs on 1GB, so it should be ~56MB per DNG... Then looking at the D7100 specs you can see they listed the lossless compressed 14-bit NEFs as 28.5MB each, so there should be ~287 on 8GB card and they say it's only 148 for that capacity... 8GB divided by 148 NEFs gives you ~55MB per NEF...

[Ayoh]

I was sharing your optimism regarding the implementation of mechanical stabilization in video until i saw some sample K3 footage which shows the characteristics field of view crop, and exaggerated jello effect of the bad implementation of pentax software SR. I find this really frustrating as it is obvious the software "movie SR" mode is seriously flawed. It seems the Pentax engineers responsible for this are not competent. The reviewers who have had a chance to use the K-3 camera so far also leave much to be desired in their rigour, as no thorough insight into the video implementation has been presented.

See details below:

https://www.pentaxforums.com/forums/pentax-k-3/239927-k-3-video-shake-reducti...ml#post2544520

Yes, if things were different.

But both menu items designate stabilization during video. It is just not meaningful to deduce implementation from this.

It is like saying: the K50 has a burst mode menu item, so the K-3's burst mode must be same speed as K50 because they named it the same. It simply isn't a possible deduction. Same for "Movie SR".

[falconeye]

i saw some sample K3 footage which shows the characteristics field of view crop, and exaggerated jello effect

Ok, now this is evidence. Thanks for the info.

And yes, IMHO it is a rather disappointing aspect if the information proves correct.

[Ayoh]

There are also a number of additional sample videos available below confirming software based SR.

http://www.onfoto.ru/review/Pentax-K-3/2520.html

The videos inside the church and on the riverfront are particularly demonstrative of the problems with the software SR implementation; image skewing, wobbling etc. I suspect they are using a Milbeaut processor function (or their own implementation) designed for global shutters and devoid of any compensation for rolling shutter exposure, hence the exaggerated wobbling artefacts.

Admittedly these are from a pre-production unit and show some obvious skipping artefacts suggestion beat firmware. However the fact that the same software based SR solution made it into the penultimate firmware is not encouraging.

Disappointing indeed.

Ok, now this is evidence. Thanks for the info.

And yes, IMHO it is a rather disappointing aspect if the information proves correct.

[Class A]

The videos inside the church and on the riverfront are particularly demonstrative of the problems with the software SR implementation; image skewing, wobbling etc.

I saw the riverfront video and the wobbling of the tower tops indeed looks ridiculous.

However, how do you suggest is this caused by software SR as opposed to just rolling shutter?

It is possible that the stabilisation characteristics of the hardware SR are preferable w.r.t the movement of the whole image , but I don't see how different stabilisation approaches could create different wobbling, etc. within the image.

Are you speculating that the software SR tries to improve some "within image" artefacts and makes things worse because it isn't tuned to the particular sensor read-out characteristics of the K-3?

[Ayoh]

I think the software SR function is probably a feature of the Milbeaut processor which Pentax uses, and that it was designed with global shutter sensors in mind and does not accommodate time lag between the top and bottom of the video image introduced by rolling shutters. This lag exaggerates the skewing effects. Perhaps if the rolling shutter read speed was much faster than the typical ~1/30 sec used by Pentax than the software SR would function acceptably. However it would still need to crop the sensor view to work (by the way on the k30/50 series the crop remains even with movie SR off!). Either way it is a really half baked design solution and should have never made it to the market. Why they continue to use it considering how badly it affects video quality is baffling, not to mention frustrating. At his stage I am thinking that the k5 series will offer better overall video quality than the k3.

I saw the riverfront video and the wobbling of the tower tops indeed looks ridiculous.

However, how do you suggest is this caused by software SR as opposed to just rolling shutter?

It is possible that the stabilisation characteristics of the hardware SR are preferable w.r.t the movement of the whole image , but I don't see how different stabilisation approaches could create different wobbling, etc. within the image.

Are you speculating that the software SR tries to improve some "within image" artefacts and makes things worse because it isn't tuned to the particular sensor read-out characteristics of the K-3?

[JimmyDranox]

I saw those movies, and the sound of AA simulator can be cleary heard, when the focal length is changed. And a jump on the image is the consequence. But those movies was made with a preproduction camera. Probably, this problem will be fixed in the series production. Must be.

[Class A]

I think the software SR function is probably a feature of the Milbeaut processor which Pentax uses, and that it was designed with global shutter sensors in mind and does not accommodate time lag between the top and bottom of the video image introduced by rolling shutters.

That may be the case but why would the hardware-based SR show any less rolling shutter effect?

I understand that you object to the image being cropped for SR purposes. But I don't see why the hardward-based SR would alleviate any of the issues you mentioned if they are rolling shutter induced.

EDIT: I just watched the church movie and I now think I know what you mean. It seems like the software SR sometimes tries to correct for the rolling shutter effect (taking the wrong cues for image stabilisation), leading to an overall jarring experience. I think your statement is that if the rolling shutter effect were left alone (i.e., no ill-conceived attempt at correcting it were made) then its impact on IQ would be acceptable.

I too cannot believe that this pre-production behaviour will make it into the final firmware. It does not make sense to add a headphone connector to a camera and then cripple the video IQ to unacceptable levels.