[Giklab]

view finder light pollution?

Probably not in video, as the mirror is locked up.

[K David]

view finder light pollution?

I'd be surprised by that. The K-3 mirror makes a reasonably tight seal and the individual frame exposures for video are quick enough that the amount of light coming in through the viewfinder should be negligible. Typically, viewfinder light pollution is an issue with long-exposure photography like astronomical work. To illustrate that, I took a number of four-minute overnight exposures with my K-3 a couple months back. I didn't have a viewfinder shade and the hotel behind the camera was significantly brighter than the star field in front of it. There wasn't any noticeable light leakage or contrast loss in any exposures.

[Al_Kahollick]

I think part of the problem is Pentax seemingly only has video capabilities because "everyone else" has it. That said, I think it's important to remember your shooting video with a camera that's been designed to deliver outstanding still shots, which it does in spades! Looking for "camcorder quality" from a still camera is, I think, asking too much.


But that's just my opinion, I could be wrong.

[stevebrot]

The K3 - and for the most part - any dSLR requires a cinema technique more than a "family-cam-corder" technique.

Well, understood!

My concern is that two clips out of the approximately 20 or so that I shot had this problem (white outlining) with no apparent change in subject, lighting, or technique. The rest were quite satisfactory despite my clumsy approach. I will be in conversation with Ricoh/Pentax regarding the issue. If it is a sketchy sensor, I prefer to have it fixed rather than have it fail in regular (still) shooting.


Steve

[Steve.Ledger]

While there is no actual EXIF for .mov files, there is a fair amount of Maker meta data in the file that can be extracted with EXIFtool.
According to the data there's no excessive sharpening or effects applied.
--
Metering Mode : Multi-segment
White Balance : Auto
Focal Length : 8.0 mm
Saturation : Normal
Contrast : Normal
Sharpness : Normal
DSP Firmware Version : 1.11.28.18
Picture Mode : Video (Auto Aperture); 1/3 EV steps
Drive Mode : Video; n/a; Shutter Button; Video
Lens Type : A Series Lens
Image Tone : Natural
Shake Reduction : Off (AA simulation off)
Hue : Normal
White Balance Auto Adjustment : Off
Tungsten AWB : Subtle Correction
Dynamic Range Expansion : Off; 0; 0; 0
High/Low Key Adj : 0
Contrast Highlight : 0
Contrast Shadow : 0
Contrast Highlight/Shadow Adj : Off
Monochrome Filter Effect : None
Monochrome Toning : None
Shadow Correction : Off
Bleach Bypass Toning : Off
AE Exposure Time : 1/144
AE Aperture : 4.0
AE ISO : 100
AE Max Aperture : 3.4
AE Max Aperture 2 : 3.4
AE Min Aperture : 23
AE Min Exposure Time : 1/7723
Lens Type : A Series Lens
Auto Aperture : On
--

To my eye it's definitely the Focus Peaking overlay which has been saved [merged] over the video file. You can see the white highlighting disappears as the balloons get closer to the camera (and therefore out of focus). This must surely be a random bug.

@Al_Kahollick

Looking for "camcorder quality" from a still camera is, I think, asking too much.

But it's not a still camera, modern DSLRs are actually manufactured to deliver stills and video. Also worth noting is that video is actually 24 stills per second at 1920x1080p. The notion that a DSLR is not a video recorder obviously only applies to earlier models where video recording was not a feature. You don't have to use it, just as you don't have to use the intervalometer, the built-in flash or any other feature you personally find irrelevant. It's a choice each of us can make with the gear we buy.
It would be impossible to get Camcorder style video from an APS-C sensor with DLSR lenses.
Rather than asking too much, I think getting camcorder IQ from a VDSLR would be a downgrade. They are really different tools which produce different results.

[Tanzer]

One thing I wanted to mention, in case nobody noticed it, is that the white highlighting is oriented along the vertical edges in the image. This is probably most obvious on the curved letters on the balloon. While I don't use live view often on my Pentax cameras, that is typical of focus peaking implementations, whereas sharpening would be on all edges, regardless of orientation.

Also the highlighting kind of "pulses" on and off in the video as the camera moves around. This is typical of focus peaking, as the image processing equates motion blur to out-of-focus, but everything becomes sharp again when the camera momentarily stops moving.

So I'd say it is definitely the mysterious focus peaking bug that was previously reported on the forum.

It would be nice to be able to reliably reproduce the problem.

[stevebrot]

While there is no actual EXIF for .mov files, there is a fair amount of Maker meta data in the file that can be extracted with EXIFtool.
According to the data there's no excessive sharpening or effects applied.

This must surely be a random bug.

Yes, probably a bug. The only causative factor I could think of might the temperature.

So I'd say it is definitely the mysterious focus peaking bug that was previously reported on the forum.

It would be nice to be able to reliably reproduce the problem.

My thoughts as well.



Steve

[wombat2go]

Playing that Dropbox .mov file on firefox with flash, it looks a bit like de-interlacing problem.
(Or maybe de-interlacing when not supposed to?)

[luftfluss]

Yes, probably a bug. The only causative factor I could think of might the temperature.




My thoughts as well.



Steve

I just turned on ESPN, they were showing highlights of the MLB All-Star Game. The initial video clips they showed - all from the centerfield camera - had the same kind of outlining as in your OP.

[PiDicus Rex]

Stevebrot : Can you copy the file back to the SD card, and open it in the camera, an cycle through to the detailed Info setting?
I'd hazard a guess that the Sharpening/Fine Sharpening setting is wound all the way up on the profile that the vision was shot in.
(I'm unfamiliar with the discussed lens, but would not expect that sort of high-contrast error to come from the glass)

[stevebrot]

I'd hazard a guess that the Sharpening/Fine Sharpening setting is wound all the way up on the profile that the vision was shot in.

As mentioned above, normal appearing video and still shots were taken immediately before and after the clip. Camera temperature was 36 C (97 F), but still below than the 40 C maximum temperature stated in the user manual.


Steve

[kadajawi]

It looks like too much sharpening, amongst other things. Perhaps it was really a bug, can we see screengrabs from the videos that turned out fine? It would be really interesting to see if those are completely different shots, or similar to this one.

Could it be compression artifacts? It does look a bit like that. The scene has A TON of detail, and combining the not so efficient/high quality encoder of the K-3 with the low bitrates it creates and the details (and you didn't even turn down sharpening), the result will be a mess.

You have to set the sharpening to the lowest possible value... sharpening only creates fake details that the encoder then also has to squeeze in, making the job even harder, and increasing the bitrates necessary to get artifact free videos. Just do the sharpening in post, make sure no bit is wasted.

Heat has never been a problem for me, apart from hot pixels. 2 days ago my camera got so hot that it overheated in live view, but recording video was fine.

[stevebrot]

Could it be compression artifacts? It does look a bit like that. The scene has A TON of detail, and combining the not so efficient/high quality encoder of the K-3 with the low bitrates it creates and the details (and you didn't even turn down sharpening), the result will be a mess.

You have to set the sharpening to the lowest possible value... sharpening only creates fake details that the encoder then also has to squeeze in, making the job even harder, and increasing the bitrates necessary to get artifact free videos. Just do the sharpening in post, make sure no bit is wasted.

Thanks for your observations. Every little bit is appreciated. See above posts and comments. Average bitrate was 19.5 Mbps with sharpening, contrast, and saturation set to default values. Here is a download link to the same video if you wish to do forensics or play using other than Flash:

https://www.dropbox.com/s/dv3bquy2y4jzr6q/IMGP0716.MOV?dl=1

I will post a link to a similar video without artifact taken a few minutes later of the same subject.


Steve

[stevebrot]

can we see screengrabs from the videos that turned out fine?

I will post a link to a similar video without artifact taken a few minutes later of the same subject.

OK...Here is a clip made two minutes before of the same subject.

https://www.dropbox.com/s/1h3xzyy2ad05dt9/IMGP0715.MOV?dl=0 (Flash)

https://www.dropbox.com/s/1h3xzyy2ad05dt9/IMGP0715.MOV?dl=1 (Download .mov)

Steve

[richandfleur]

So I'd say it is definitely the mysterious focus peaking bug that was previously reported on the forum.

Has this been reported elsewhere?


It's definitely bad, and whilst sharpening could start to produce this level of edging artefact, this does look more like the focus peaking 'filter' that is normally sent to the rear screen prior to recording.
It looks like in this case that filter was not switched off, and the effect was stored to the card.
Do you normally use focus peaking?


The way Pentax do this is disappointing, as other brands display the focus peaked video on the rear screen, whilst simultaneously writing the non peaked file to the card.
I've not been super impressed by the quality of Pentax software, (the intervalometer just stopping is another example), and this is the type of thing they should be investigating and correcting quickly.
Hopefully they are able to identify where it's going wrong, which is the hard part of software fault finding.


Initial feeling is the sensor is ok here. If it works on other videos and your stills are fine, then this is likely a software bug.