[Filter Test SEU Gen2 #9A] Longpass Stack Wandering Discussion. See #9B for results.

[Cadmium]

You can do all that, with a longer exposure or tweaking the RAW to boost exposure later, but why?

It might be interesting to compare over exposed version of different filters to see which one has the strongest hidden tendency to leak.

But this is not about that.

All we want to see in the stacked image is everything minus the UV.

If there is nothing there to be seen using the same exposure, that is all we need to know.

 

I guess you can up the exposure in RAW, but that doesn't show reality.

[Andy Perrin]

OK, I just gave you two examples of that up in Post #21.

But can you tell me why it makes sense to use the same exposure for two filters that was used for the single filter?

Assuming by "two filters" you mean the filter+LP vs. filter alone, because (I thought) you were trying to tell if there is any contamination in the latter? Blocking all of the UV out leaves just the contamination, so if it's black at that same exposure and processing, then the image is uncontaminated.

 

I try so hard to write carefully, but somehow I manage to goof it up all the time....Or maybe I write so carefully that it is utterly boring and stuff gets skipped over.

People read these forums in a pretty non-linear way. When you click, it jumps to the end of the thread, and if you're like me, half the time you read the end first then go back to the start if it sounds interesting. But that out-of-orderness sometimes leads one astray.

[Cadmium]

Andy gets it.

 

As far as Raw Digger, your on your own, I don't have it.

My JPG and RAW images look the same from the camera unless processed in some way.

I can't discus your experience with Raw Digger. Maybe someone else.

 

Here is a redone version of my Baader U, and Baader U + RG610 comparison.

These are all the same white balance from camera. No processing, except the noted.

Settings are noted, all the same except exposure. Shot on a sunny day.

Bottom right pic has added exposure compensation in RAW, but keep in mind, this means nothing, it isn't real, the left hand black image is real,

real like the UV shot above it, same 2 second exposure time.

Both right hand images are not real, they are both pushed... Interesting in some ways, but not reality, and would be confusing to anyone looking at them that didn't realize they are not real.

Those right hand images do not represent the amount of out-of-band 610nm and above that is present in the top left UV-only shot.

The bottom left hand black image is representative of the amount of out-of-band 610nm and above that is present in the top right UV image, none.

 

[ulf]

To me every step Andrea has made is completely logical and I do not understand the argumentation against it.

That might be because I never use anything that is just automatically processed to a jpeg.

 

Such processes in common programs almost always include clipping into the darkest and brightest areas.

That is over-processing, dumping valuable image data.

 

When doing a proper controlled processing of the RAW image sometimes what is lurking in the black basement jump up and bite you.

I sometimes see problems with leakage in BaaderU images, even if that filter is considered as OK here from leakage point of view.

[Cadmium]

Ulf, Why is it logical to you to use any other exposure time for the UV cutting stack shot than is used for the UV only shot?

[Andy Perrin]

I never use the camera JPEGs either, for that reason. But I still expose the photo normally, so the light meter is reading 0.7 to +1 or so, then I convert in PN. Not sure that affects the test much, except for the difference between 12bit and 8bit graphics. If quantization error matters then you are truly in the weeds!

 

When doing a proper controlled processing of the RAW image sometimes what is lurking in the black basement jump up and bite you.

I sometimes see problems with leakage in BaaderU images, even if that filter is considered as OK here from leakage point of view.

Well yeah, but in those situations people are almost always doing something other than shooting under normal sunlit conditions. It seems like almost any filter or stack is going to have some leakage which you can see if you push it to the max. I guess we could rate relative levels of such leakage among different filters/stacks? In that case it's clear that the U360 + S8612 stack beats the Baader handily by this measure.

[Cadmium]

My question is, why would you use a different exposure for the stack?

[ulf]

Because in these tests Andrea is trying to scientifically analyse the properties of the filters and get a good comparable view of the leakage.

This is very much the way one change the time doing spectrometric analyses too, to avoid nonlinearity effects of the sensor and see the leakage clearly.

It is a more complex method, but it absolutely makes sense to me.

 

Then if the leakage will be a problem in practical use depends very much of RAW-processing, the light situation and subjects to be photographed.

I have seen leakage problems with my BaaderU for some flowers that are very UV-black and IR-White.

Your scene above is not fully representative of all possible image situations.

[Cadmium]

What I am doing is to scientifically analyze the properties of the filters and get a good comparable view of the leakage.

Andreas is changing the ratio by changing the exposure time.

It is interesting to look at the longer exposure to see beneath the ground, so to speak, somewhat like how a diabatic graph shows more...

However, a longer exposure time for the stack doesn't show you the real ratio of out-of-band to UV that is present in the UV shot.

The out-of-band that is present in the UV shot is not strong enough to show up, because at that exposure time, the UV is stronger, unless of course the out-of-band shows up at the same exposure time alone,

then it will be polluting the UV shot.

[Andy Perrin]

The out-of-band that is present in the UV shot is not strong enough to show up, because at that exposure time, the UV is stronger, unless of course the out-of-band shows up at the same exposure time alone,

then it will be polluting the UV shot.

What about Ulf's comment about seeing it in the dark areas of the shot where there's less UV to compete? I guess that matters for the flower photos. (It does seem like the Baader actually has the worst problem with this, however.)

 

Maybe some regimen like:

 

-use a flower with UV dark spot (dandelion? daisy?)

-test Baader, U360+S8612, and SE2 and find out what a normal exposure is for those.

-retest with GG420 on each at that same exposure and analyze with Raw Digger

-ditch the "as shot" photos which none of us use anyhow (what do we need those for?)

-determine which has the most leakage

-repeat for a different lighting scenario, like flash or something

[Cadmium]

If any out-of-band shows up in the UV blocking stacked shot, then it will show up in dark areas of the UV shot, yes for sure, then you have a leak. Yes, that matters...

If nothing shows up in the UV blocking stack shot then nothing shows up in the dark areas of the UV shot.

Go back to Andrea's post:

http://www.ultraviol...dpost__p__22741

In that she shows a black frame, then an altered version of the black frame. I don't know the process of that, if it or all were Raw Digger or some other program.

If you like the results of Raw Digger, then use it, then base your insight on those results, but I still think the exposure time should be the same for the UV blocking stack shot, yes,

because that compares the true ratio of UV + Out-of-Band to Out-of-Band alone. They should both be shot using the same exposure for a real comparison.

[Andrea B.]

However, a longer exposure time for the stack doesn't show you the real ratio of out-of-band to UV that is present in the UV shot.

 

I am completely flummoxed as to why you fail to understand that in Post #10 I was NOT LOOKING FOR A RATIO OF OUT-OF-BAND TO UV that is present in the UV shot. I was trying to trap that d@mn3d green bump. Nothing more than that. Go BACK and RE-REaD Post #10. I'm starting to get weirded out by this.

 

*******

 

And here is where I would like to eventually go: Raw Digger gives you a count of the actual raw data in the file at each intensity level. So you could take the Red, Green and Blue channel counts at each intensity in the UV-only photo and compare it to the R, G and B channel counts at each intensity in the UV+Longpass photo (at the SAME exposure time) and perhaps come up with an actual measure of how much contamination you have. I would try this if we all could ever get off high center on this green bump trap thing.

 

For one dollar and a cup of coffee I would pull this entire topic and start over.

[Andrea B.]

OK, I'm off to Walmart this morning over in Ellsworth to pick up a backup SD card. There won't be any interesting photography today because the summer fog which has come right up to the edge of the deck and an overcast sky are not providing much light of any kind.

 

I'll get back to this topic and the debate in due course.

[dabateman]

Andrea,

The most important question was never answered, what movie did you watch?

 

Joking aside, I think Andrea did an excellent job trying to nail down the issue of if that 524nm band is a major concern for the SEUmk2. Short answer is no.

Long answer will seem to depend on the light source and imaging conditions.

To really stress test a baader or other filters I use a Hallogen 72W bulb. It does output uv. I see a green band in Baader filtered 1000lines/mm test. However, its ratio of uv to IR is heavily skewed to the IR making it the best stress test light source.

Andrea if weather is bad and you want hard stress test for filters. See if you can buy the GE hallogen bulb, its a halogen light inside a regular A19 bulb allowing you to just screw it into a desk lamp. The hallogen inner bulb is not uv filtered like other hallogen bulbs that fit into specific hallogen fixtures. They are also cheaper, $5 at Target for clear glass 72W pack.

To stress test the SEUmk2 the only other bulb I can think of is a sodium bulb. It has hard lines at 588nm. But I don't think anyone uses sodium light, so only source maybe Home depot. Also since no one uses them the results are probably meaningless.

[ulf]

Andrea, I'm sorry if I helped to let this topic go astray.

 

Below is what I had written when the topic became locked after your post #37 and was unable to post.

I think that is still has some relevance.

 

----

This thread started to determine the level of "leakage" from the green peak at 524nm for the SEU2, not how much that peak impact the final image.

All filters has a limit to their blocking density and out of band light can be forced through with enough time or light intensity.

The linear output from Raw Digger show what the sensor really see.

 

All Raw converters producing a "normal image" do a nonlinear conversion of the original information to make the picture look close to how we see the scene.

That nonlinear transformation is not identical between Raw converters.

By including a specific scene, illumination situation and Raw-converter you add much unknown factors. Then it is only possible to draw conclusions from that limited set of inputs.

 

IMHO it is best to first determine the level of leakage and then see if it will be a problem for the type of photography and processing the filter will be used.

In many cases a BaaderU will be quite sufficient and possibly the easiest alternative for pure UV. It was for me, until I found the leakage problem in some images.

Now I often prefer the U-360 S8612 stack and take the penalty of almost the double exposure time.

[Andrea B.]

The most important question was never answered, what movie did you watch?

 

Annihilation

[Andrea B.]

Ulf, please see the new topic where I compare the raw data histograms of a UV-only photo and its companion UV+Longpass photo in order to determine how possible IR contamination affects the UV-only photo. I'm not sure I have it correct just yet, but I think it is a promising area to look into. All suggestions and comments and help will be welcomed because I would like to find a quantitative method which could be used by everyone to determine this. LINK: http://www.ultraviol...iff-might-work/

 

Added 2018.07.25: Ignore that histogram experiment. It failed big time!! :rolleyes:

[Cadmium]

The test I suggested for out-of-band leaks requires the same exposure time for the UV, and UV suppressed shots,

and the test works good for a typical UV exposure time of say 2s, or there about.

Here is the flaw in my thinking.

In some situations I have had to use longer UV exposure times, say 8s, 10s, 15s... etc., which is a bit long.

In a situation like that, the UV suppressed shot would need to be the same exposure, and that might start to show more out-of-band image than the typical 2s pair I illustrated.

Now, such scenarios that need more UV exposure also may not have the usual UV to out-of-band ratio encountered in a more typical 2 second exposure of a flower on a nice day.

There obviously isn't as much UV available in a UV shot that requires a longer exposure, but there might be just as much visual and IR or even more in some situations than the flower shot on a nice day.

So a different ratio of UV to out-of-band light, and a longer exposure, so not only might the scene have a higher ratio of visual and IR compared to UV, but the exposure time is longer,

so this would create a much more out-of-band weighted situation, and we might expect to see more out-of-band pollution in dark areas.

Unless of course, the out-of-band light was minimal.

 

Then we have some very extreme UV situations, such as UV night shots, which by nature usually have very little UV, and more visual and IR light.

Most night time lighting is designed to emit visual, and often has a strong IR content, and UV is present in some lighting, but I imagine most night lighting has a lower ratio of UV to visual and IR light.

Such exposures can require minutes, not just seconds.

That is the extreme, and we might expect to see a lot of out-of-band leaking with the usual UV filter when exposures are minutes.

 

However, even with exposure times of just 8 seconds, and even assuming the UV to out-of-band light remains the same as the 2s scenario (and it probably doesn't),

we have increased the exposure 4 times, so we will probably start to see more out-of-band pollution in the dark areas.

 

Sorry, my thinking was only half way correct.

One could still do a test shot for a long exposure scenario, it would not be a bad idea at all, but yes, the same exposure time shows the actual amount of out-of-band light.

 

So the flaw is this: If we use the test for a UV exposure time of 2s, and we see no image in the 2 seconds UV suppressed image, this is good, but only good for that specific scenario and that UV exposure time,

because the test falls apart with longer UV exposures, and with such situations one might want to run such a test to know how much out-of-band light there is.

The test is specific to the exposure time it is used for, per filter, the results for a 2s UV shot can not be relied on for longer UV exposures using the same filter.

 

With longer exposures, it might be more important to use UV filters with stronger Red/IR OD suppression.

Stronger out-of-band OD often also means lower UV peak transmission, so finding the best optimized ratio between UV peak/bandwidth and red/IR suppression might be more important with longer and unusual UV exposures.

[Andrea B.]

Good reasoning! Thank you for the follow-up post, Cadmium

 

---

 

I would like to thank everyone for all the comments in this topic.

 

---

 

The Longpass tests still need to be done and will be. But I think I will start a new topic for the results.