[pres589]

4 to 5 stops??? Definitely show off some samples doing the comparison.


That's the kind of change that makes me want to get another K-5 II or IIs and have this done.

[Medex]

4 to 5 stops??? Definitely show off some samples doing the comparison.


That's the kind of change that makes me want to get another K-5 II or IIs and have this done.

For the sake to not overestimate the improvement I added RAW files to the google disk. The link is below. Download files and analyze them in your own way.
Pictures were taken in these conditiones:
1. Printed test chart fixed on the wall (horizontal line is checked with spirit level)
2. Tripod, Sigma Art 18-35 lens at f2.8 and 35 mm, the distance from the test chart to the sensor 4x focal length
3. Camera aligned horizontally as much as possible according to horizontal line in the test chart
4. 12 s timer
5. Continuous lighting
6. LiveView autofocus for the first image, then AF switched off (K5ii normal pictures were autofocused / checked +3 times (aditional pictures), but they do not differ - LV autofocus is right in all cases)
7. Exposure was set in LiveView to match histograms between K5 mono and K5ii normal - the difference is about 2/3 of stop - monochrome sensor catches more light
8. ISOs: 100 - 400 - 800 - 1600 - 3200 - 6400 - 12800 - 25600 - 51200

Postprocessing: All monochrome pictures are converted in Monochrome2DNG converter to skip demosaicing algorithm in LR raw converter and get maximal resolution.
Color pictures can be monochromatized by applying Embeded BW profile or -100 saturation.

Just for curiosity try Luminance noise reduction slider in LR and compare details between ISO6400 mono and ISO100 color pictures (6 stop difference).

Update your browser to use Google Drive, Docs, Sheets, Sites, Slides, and Forms - Google Drive Help

With monochrome sensor you get pixel shift resolution handheld, just monochromatic.

[Michail_P]

Better resolution is expected.. no color noise. Your work is brave. Congrats on the results and the effort.

[Medex]

Better resolution is expected.. no color noise. Your work is brave. Congrats on the results and the effort.

Thank you

[BenwayB]

Hello,

ok!

This threat did relly make me want such a conversion, too :-)

Yersterday i purchesed an old K110d, today i did my best to make the conversion.

// This was the first time, that i opend the housing of a DSLR ;-)

Everything went pretty smooth, as i was able to follow the detailed service manual.

As the sensor has been removed, the real challange began ;-)

* There was the IR filter (screwed) and an additional glas in front of the sensor (glued...). To remove the second glas, was no fun.
* I started with variuos solvents, but nothing worked. So i started with wood and ended with brass. From my opinion, i achieved a very good surface.

After re-assambly, the camera does start, does AF and does fire ... but than i freezes.
Ater turning it off and on again, i can do the same thing again.
Live-View does the same thing ...


Any suggestions?
Is it worth, to disassamble it again, or is the sensor damaged?

How sensitive is wich part of the sensor, when working on it?

Best regards :-)

[BenwayB]

And Pictures...

[BenwayB]

oh!
Acutally it DOES fire, but only without a lens attached to it ;-)

this is, what i get:

Attached Images

View Picture EXIF

File Date

 

[Medex]

oh!
Acutally it DOES fire, but only without a lens attached to it ;-)

this is, what i get:

Hard to say where you made mistake. K110d as well as k100d sensor is very sensitive to scratches near imaging area border.
Maybe some solvents left beneath the sensor plate and shorten circuitry... I don't know. But seems that sensor is not working.

[BenwayB]

Hello,

thanks for the reply :-)

How much distance do I have to leave as it is?

1mm? 0.5 mm?


Actually i did scratch almost the entire surface area, and as you can see on the first picture, right side, even on the blue area.

New cameras to be killed are orderd, already ;-)

[Medex]

Hello,

thanks for the reply :-)

How much distance do I have to leave as it is?

1mm? 0.5 mm?


Actually i did scratch almost the entire surface area, and as you can see on the first picture, right side, even on the blue area.

New cameras to be killed are orderd, already ;-)

Dry scrathing of CCD near blue area poses risk of damage. It is my opinion. Not proven. So scratching should be very gentle here. I had 3 failures with k200d CCD sensor and think the problem was some minor damage od circuitry on the border between imaging area and blue area.
Newer sensors like CMOS in k5 are not so sensitive here but they require even more gentle scratching due to scratch sensitive passivation layer

[BenwayB]

Hello,

I decided TO start with the k110 due to the fixed sensor.

To brake the glass of the sensor, has been very difficult, but besides that, everything was easy.

How much more difficult is it, to work on a SR Sensor?

How much harder is it, to work on the K10 compared to the K110?

Many thanks and best regards :-)

[Medex]

Hello,

I decided TO start with the k110 due to the fixed sensor.

To brake the glass of the sensor, has been very difficult, but besides that, everything was easy.

How much more difficult is it, to work on a SR Sensor?



How much harder is it, to work on the K10 compared to the K110?

Many thanks and best regards :-)

K10d presumsbly has same sensor as k200d. CCD. It is very scratch sensitive.
If you want work with the sensor that has SR system, it is better to remove sensor plate from it. Be careful snd dont loose small metal balls.

[BenwayB]

Hello,

today I did it again, and I succeeded!
I have been a lot more careful to stay away from the edges. Also I did not mess around with solvents.

Still no 100% but I'm getting closer.

This time the sensor works, but has a vertical line that is dead.

I also struggle with the raw conversion, as my pictures show a visible pattern.

Next time I think I will try it with solvents again, as i have access to things like THF, DMSO, Chloroform and so on, and access to a lab where I can handle it save.

I expect solvents to be less stressful for the sensors :-)

---------- Post added 01-28-22 at 03:13 PM ----------

This is a sample Image

---------- Post added 01-28-22 at 03:20 PM ----------

I think I did not get, how to make the raw-conversion correctly.

I start with a raw file, that has the file format *.PEF.
This image ist coloured on the edges and b/w in the center.

Then i used "monochorme2DNG"

The result is a SNG-file, that is sligtly larger in size (by pixels), but STILL with colour !!

When i import it to LR, lightroom identifies it as monochrome ...

??

Please help me a little bit, i think i don't get it :-)

[BenwayB]

Hello,

ok, I figured out that the converter does maintain a coloured thumbnail, so things make sense :-)

Now I have one week to test the camera.

[acoufap]

Hello,

today I did it again, and I succeeded!
I have been a lot more careful to stay away from the edges. Also I did not mess around with solvents.

Still no 100% but I'm getting closer.

This time the sensor works, but has a vertical line that is dead.

I also struggle with the raw conversion, as my pictures show a visible pattern.

Next time I think I will try it with solvents again, as i have access to things like THF, DMSO, Chloroform and so on, and access to a lab where I can handle it save.

I expect solvents to be less stressful for the sensors :-)

---------- Post added 01-28-22 at 03:13 PM ----------

This is a sample Image

---------- Post added 01-28-22 at 03:20 PM ----------

I think I did not get, how to make the raw-conversion correctly.

I start with a raw file, that has the file format *.PEF.
This image ist coloured on the edges and b/w in the center.

Then i used "monochorme2DNG"

The result is a SNG-file, that is sligtly larger in size (by pixels), but STILL with colour !!

When i import it to LR, lightroom identifies it as monochrome ...

??

Please help me a little bit, i think i don't get it :-)

Raw Therapee offers a lot of demosaicing methods (=> raw tab) including "mono" and "none (shows sensor pattern)" to create an image from a raw file - maybe in your context it's worth having a look at the different approaches.