Camera sensitivity - chasing ghosts in spectral sensitivity measurements

[Andrea B.]

Do you know why the anti-aliasing filter was not removed the first time?? I thought they were always supposed to be removed when converting a camera?? Just curious. :D

 

---

 

I've had people misinterpret your sensitivity charts as a transmission chart. Sensitivity might be low at 340nm, for example, but you can still shoot there fairly easily with the appropriate filter and adequate UV light.

 

 

 

.

[JMC]

Andrea. I hadn't realised that the 5dsr has two optical low pass filters. In the original conversion one was removed as part of the IR blocker assembly. The other was left on the sensor. This apparently is common practice during the conversions (I have no idea whether it is or not, but am in no place to doubt that). It is the last line of protection for the sensor, and trying to remove it, more often than not leads to sensor damage. During the latest visit, this second one was removed. In the monochrome conversion it is automatically removed, as everything needs to be to get at the sensor.

 

The sensitivity charts are a way to try and understand relative sensitivity across a range of wavelengths. With enough light or long enough exposures then yes, imaging down at 340nm is perfectly feasible.

[Andy Perrin]

I just got a 340nm bandpass filter today, and even with my normal conversion, 340nm isn't particularly hard. I was using 15sec ISO200 F5.6 on my Sony A7S.

[Andrea B.]

Thanks, Jonathan, for the explanation. It takes two layers to anti-alias, but I had not realized one was stuck to the sensor. Many recent cameras don't have anti-aliasing anymore. I'm wondering if I should make this point in the Sticky and encourage people to convert cameras without anti-aliasing?

 

******

 

Andy, what filter did you get?

[Andrea B.]

Wait a minute......if you only remove ONE of the anti-aliasing stack, won't your photos be blurry?

[JMC]

No idea Andrea. I couldn't notice any particular blurring with only one removed.

[Andy Perrin]

Andy, what filter did you get?

Omega's "Optical Filter 340BP10 Excite FURA 20mm"

Not the greatest filter in the world, but it seems to work. At least, the results look like everyone else's 340nm pics.

[Andy Perrin]

The image should get *sharper* when you remove an anti-aliasing filter. The anti-aliasing deliberately blurs out the high frequencies in the image that are too high to be properly displayed by the sensor (leading to moire patterns, also known as aliasing).

[Andrea B.]

I was trying to say that if the antialiasing is done with two layers, one for the vertical and one for the horizontal, then removing only one layer might be weird.

[Andrea B.]

diagram showing what I did not say clearly !!

The layer in Jonathan's camera corresponding to the green layer in this diagram was apparently glued to the sensor and not originally removed? So you would still have some blur?

 

[Andy Perrin]

Yeah, you would get blur in one direction but not the other I guess?

[dabateman]

If Jonathan has some standard object that he photographed previous to conversion and now after, he could answer that question. It would need to have horizontal, vertical and diagonal lines to rally know. A test chart would be best.

 

Jonathan,

The sensitivity seems odd less than 340nm. You have blue being high, but I don't think we see that. Red yes a little. But its mostly green less than 340nm. Is your background noise really high?

See my last post here:

https://www.ultravioletphotography.com/content/index.php/topic/3229-freeware-em1-dye-channel-images-using-bandpass-filters/

No blue, mostly green, little red.

 

[JMC]

David, I find it strange about the green too, and I am not sure the reason why. It could be down to the very low levels of signal in the short wavelengths. I'm not sure how the greens are processed - there are 2 green channels. Are they added together, or averaged? I don't know. If they are added together that would explain the green, but it doesn't seem right to me. At the end of the day, the pictures are the definitive proof. It looks green in the image at less than 340nm and for some reason my data does not directly reflect that.