[dave2k]

I recently shot a number of long exposure pictures of waterfalls with my K-1. After taking each picture as a JPEG I normally inspected the image and then saved the RAW image on those shots that looked like they might be enlargement-worthy. I thought the RAW images would be especially useful since a considerable portion of most shots was relatively dark as I exposed for the bright water. I've been reviewing the shots in Photoshop Elements 14. While inspecting the JPEGs for sharpness and focus, I noticed some brightly colored spots that are larger than pixels. After finding these spots on multiple shots, I realized most of them appear in the same locations on multiple shots. I think I found about 6 or so spots on the affected shots in a variety of colors. The spots do not appear on the RAW images. So I think this rules out a "hot pixel" type problem and instead points to a problem with the JPEG engine. But how could the JPEG engine create flaws in the same locations on multiple different shots? Or am I totally going in the wrong direction on this. The two crops below show a bright red spot and a darker purple spot in the same spot on these different images. These are crops with no other editing. As stated above, I found these defects in the same place on multiple shots. I realize these are easily removed from JPEGs in Photoshop and don't appear on the RAW images so I can workaround this problem, I'd just like to know what is going on.

[AldaCZ]

Have you tried pixel mapping ?

[dave2k]

Have you tried pixel mapping ?

No, I'm not familiar with that technique. The spots appear to be bigger than pixels and they do not appear in RAW, so I was leaning away from a pixel problem.

[erichope]

My comment may be irrelevant, but here goes. I have a comparable problem on my K100D which turned out to be batteries. I found when using alkalines or ordinary rechargeables I would get 3x3 or 4x4 pixel spots. Switching to high discharge & high capacity lithiums solved the problem. You mention long exposures: maybe using an external power source which will provide steady, ample power for the duration of the expose, process and save cycle will help your troubleshooting.

---

Last edited by erichope; 09-04-2017 at 02:17 PM.

[AldaCZ]

No, I'm not familiar with that technique. The spots appear to be bigger than pixels and they do not appear in RAW, so I was leaning away from a pixel problem.

It sumewhre in setup menu of the camera /like sensor cleaning etc/..it will try to correct this

[RKKS08]

I don't think sensor cleaning will help. Dirty spots on the sensor produce black spots on the picture just as there were dark spots in the scenery.

The situation can be different with pixels known as faulty. Because these pixels are interpolated from neighbouring pixels which are not listed as defect. I remember Pentax had a major problem with the first badge of the K20 (I think it was solved with one of the firmware updates?). The problem there occurred when horizontal borders between light and dark areas met a known defective pixel, with that pixel just on the darker side. The algorithm used only a small number of neighbouring pixels for interpolation. As only the light side, but not the dark side (value=0) could contribute to the interpolation, the missing pixel would be replaced by a red, blue, or green one. This was very visible specially with zebra-like horizontal patterns.

This also was a proof how many dead pixels are located on any sensor, which are disabled already by QC in the factory - for the 14MP sensor obviously 50-100.

But the problem of the OP is probably not related to dead pixels. Can it be seen on the camera display?

[dave2k]

Can it be seen on the camera display?

I just checked and the bright spots can be seen in replay on the camera screen. Maybe I'm missing something, but shouldn't a pixel problem be seen in the RAW file? There are no anomalies seen in the RAW files. Perhaps the defects are too small to be seen in the RAW file and then the JPEG engine does what you have described, actually creating the visible defects?

[amoringello]

Raw file processors automatically take care of those. Shoot Raw and you won't see them. :-)
If you have to shoot jpeg, it's very normal to have a few of these.

[RKKS08]

@dave2k
@moringello

If the problem is indeed caused by clusters of dead and hot pixels on the sensor (which I doubt), it could mean that the in-camera raw file display makes its own demosaicking, but the in-camera jpeg engine does not use these processed raw data and starts from scratch using the sensor data themselves.

A test may (or may not) reveal this: shoot raw only, and some pictures later (when you are sure the buffer does not hold any data of the picture) do an in-camera RAW to JPEG conversion. Investigate whether the JPEGs now look different.
And process the JPEG externally, after shooting RAW only. Does it look different?

[kh1234567890]

I would start by looking at the RAW file before demosaicing, using something like Rawdigger.

[Wheatfield]

I just checked and the bright spots can be seen in replay on the camera screen. Maybe I'm missing something, but shouldn't a pixel problem be seen in the RAW file? There are no anomalies seen in the RAW files. Perhaps the defects are too small to be seen in the RAW file and then the JPEG engine does what you have described, actually creating the visible defects?

Depending on your converter, hot pixels may be being mapped during conversion.

[dave2k]

I've had some more time to research this problem. I originally thought the problem may have something to do with long exposures and my practice of inspecting a JPEG on the camera screen and then saving a raw version of the shot. I thought this because I had never seen this problem until the few recent days when I was shooting with this technique. Now I've reviewed shots that go all the way back to when I first purchased the K-1 and I can see at least one of the bright spots on the vast majority of pictures. So the problem is not at all new and it's not related to any of my recent techniques. I closely inspected the RAW version of recent files with the bright spots, going up to 800%, and I could not see any of the abnormal spots. Perhaps this level of magnification is not adequate to see one bad pixel, and as suggested earlier, the JPEG engine is extrapolating the stuck pixel which makes it visible at 100% magnification. Since this now appears to be strictly a pixel problem I performed the pixel mapping as suggested and after reviewing a few test shots I can no longer see any bright spots. So my conclusion is that I had a number of stuck pixels, some of which existed at the time of purchase, and these just became noticeable as I was doing more pixel-peeping to evaluate shots for large prints. I'll keep monitoring my shots for recurrence and I can easily remove the spots on existing shots in post-processing as I pretty much know where to look for them. And I may shoot more raw shots as many suggest for a variety of reasons.