How many stuck pixels are to be expected on a second hand Fuji X-T4?

A few days ago, I bought a used Fuji X-T4 from MPB (condition was specified as “excellent”). Overall I’m very happy with it but now noticed that it seems to have quite a big number of bad pixels.

They are pretty visible across the area of the monitor when zooming in to 100% (this is the camera jpg):

Here is a crop of the lower area:

And about the same area from the raw file (disabled all modules in darktable and set demosaic to “passthrough” to verify that it is really coming from individual pixels):

Those bright pixels are always at the same position in different shots, so I thinks it’s a sensor pixel issue and not just noise. Interestingly they become less obvious when shooting with lens cap on (still visible when you look carefully but less bright), not sure why is that.
They also seem to disappear at higher ISO numbers but maybe that’s because some denoising kicks in?

I already ran the “pixel mapping” function of the camera but unfortunately it didn’t help (maybe because they are not fully saturated).

It’s the first time I bought a used camera, so not sure if this is a reason to send it back (I’m still within the 2 weeks period) and get an other one, or if this is just normal for an “excellent” used camera and the other would like have the same problem.

Any advice from more experienced users would be very appreciated.

I also should figure out how much of an actual problem this is in practice. When raw-processing I can probably fix it easily. At least on first glance darktable’s “hot pixels” modules seems to do a good job. So it mostly affects sooc jpgs. But maybe it’s not really a problem as long as I don’t crop them too heavily…

I am by no means a used camera expert but my Sony is about 4 years old and shows no dead pixels that I know of. Perhaps you may want to return it if the price is not suitable for the number of dead pixels you have found.

1 Like

That is probably due to hot pixel removal in camera.
Same with Canon 6D at ISO 6400 and above or R6 at ISO 12800 and above.

This should get rid of them…

That seems a lot of dead pixels. In my experience with non-Fuji cameras hot pixels become a problem with higher ISO and don’t seem an issue at low ISO. DT’s hot pixel module is excellent at detecting and fixing hot pixels.

Yeah, if they would only appear at high iso, I wouldn’t be so worried but they are already there at the base iso (160).

@Daniel_Catalina I already ran the pixel mapping twice, unfortunately with no effect.

I never noticed any pixel issues with my other cameras. I just did a comparison (same scene, roughly same exposure) with my now 9 years old Nikon D5100 and couldn’t find any bad pixels.

So probably I should send it back and hope that I get a better sample on the second try :weary:

Could you take a lenscap-on capture and post the raw? I wrote a simple hotpixels finder program and I can’t get out-of-family pixels on my cameras, want to see if i can see yours…

Here is one from the same day as the photo above. For some reason only the very bright ones are visible there.
20240327-230419-042.raf (10.9 MB)

On this more recent one, I can’t see any, though. Seems like the pixel mapping did something after all. It just didn’t catch all of them, only the brightest few.
20240329-084811-010.raf (11.7 MB)

I can still see many of them in photos without lens cap. It somehow seems like they are not really stuck (as in ‘always bright’) but are just somehow more sensitive than the others. Not sure if that makes any sense.

Don’t feel too bad. Here’s a bad pixels list from a Sigma SD1 Merrill Foveon-based DSLR’s meta-data:

BEGIN: CAMF matrix meta data (BadPixelsF20)
unsigned integer [3][159937]
x: D1
y: D0
4 253 8447 4 363 4607 4 405 4607 4 683 5119 4 737 8447 4 936 5119 4 973 8447 4 1057 16639 4 1063 16639 4 1166 4607 4 1189 4863 4 1439 8447 4 1460 8703 4 1671 16895 4 1686 8703 4 1697 8447 4 1790 8703 4 1865 4607 4 2010 4863 4 2052 8447 4 2083 8447 4 2124 8447 4 2129 5119 4 2175 8447 4 2472 4607 4 2560 8447 4 2575 8445 4 2590 8447 4 2937 4607 4 3174 16895 4 3375 16895 4 3401 8447 4 3739 4863 4
… (479711 skipped) …
END: CAMF matrix meta data

Lotsa bad 'uns out of about 15MP X3 = about 47MP. Fortunately the proprietary converter takes care of them but they do show up in RawDigger :frowning:

1 Like

Have you tried with other raw processor? I think for Fuji you can use a freeware version of capture one. Might be interesting to see if the problem is really hardware or maybe there’s a software bug somewhere…

Since they are also visible in the camera JPGs, I’m rather sure it’s a hardware issue.
For post-processing I’m not so concerned as they can be fixed there. However for things like family photos I prefer to just use the camera JPGs as much as possible.

This is kind of strange, as the camera engineer should know where they are and hid them when processing. This is the whole point of the mapping feature…
Do you have the latest firmware installed in the camera and lenses you are using?

Yes, updating the firmware was one of the first things I did when I got it (it was running a rather old version before). I think the problem is that these pixels are not fully stuck but are just somehow overreacting to light. As mentioned above, they are not really visible when shooting with lens cap on. I assume due to this the pixel mapping function of the camera doesn’t detect them.

Are they there in the same place with different lenses or at different focal lengths?

I didn’t have time yet to check. Anyway, I’m sending it back now and getting another one.
Unfortunately MPB increased prices in the meantime and I’ll have to pay the difference. Not so great in my opinion…