Analyzing scene SBR with an app

[Sirius Glass]

My spot meter died and I can't afford a new one. So I carry the light meter over to the bright area and then into the shadowed area and make measurements. Works great.
Actually after so many years of shooting in the Southwest and Rockies I can anticipate the values pretty accurately. A bit of bracketing and I'm sure to get a good negative, especially with films like TM400 that have a 14 stop straight line

Tabular grain films such as TM 400 are straighter exposure line and have less reciprocity problems than traditional grain films such as Tri-X 400. I use Tri-X because I prefer its grain, knowing that TM 400 has some better properties.

 

[Craig75]

The brightness range is of secondary importance in B&W photography compared to the exposure. Scene brightnoess range, of course, has never been a criteria for exposure of B&W negative materials.

Of course the shadows are used for exposure, based on the work of Nelson and Jones (which were subsequently incorporated into ASA and then ISO).

Does this app allow one to set the exposure based on the shadow areas? I don't expect you to fix a "K" value but the sophistocated user, that understands what is going on, should be able to set their own "K" value based on their own Zone I testing.

yes the app allows you to set exposure based on shadow area - it tells you the maximum and minimum f stop stop in the range so it is just 1/iso + your personal choice of 3, or 4 stops past the widest aperture the app gives you.

 

[radiant]

Does this app allow one to set the exposure based on the shadow areas? I don't expect you to fix a "K" value but the sophistocated user, that understands what is going on, should be able to set their own "K" value based on their own Zone I testing.

These apps works so that the luminosity is calculated from the pixels. Combined with exposure information you can get the EV for that selected group of pixels. Traditional mobile light meters use the image as a whole and exposure is calculated from all the pixels.

Of course by subtracting enough stops you can guesstimate shadow exposures, but the calculation provides the "mid gray" exposure.

think you will have to use hdr function to eliminate specular highlights from the reading.

Good idea, however I don't know how to implement that in sensible way as this works with "live" image. I need to remember this as a choice if needed!

Is that 15.7 the brightest area with detail or pure white? I

It is "correct" expsure for the highlights. I believe if that area was clear sky,it would be exposed to zone V.

When looking at the values of the app, remember all values are exposure suggestions for the selected group of pixels. There is also a "expose at" which calculates all the pixels together - averaging the scene.

'd also recommend working on basing the exposure off a shadow index as ic-racer suggested

I didn't quite understand the idea or my app is doing that at the moment. As it shows the "dimmest" group of pixels in terms of correct exposure, isn't this what is wanted or can you elaborate this in perspective what the app currently does? This is actually why I posted here because I'm not that advanced in exposure measurement.

 

[radiant]

yes the app allows you to set exposure based on shadow area - it tells you the maximum and minimum f stop stop in the range so it is just 1/iso + your personal choice of 3, or 4 stops past the widest aperture the app gives you.

Yes this is what I believe it does too.

If you expose based on measurement at the "dimmest" block that area becomes "correctly" exposed. However what happens to your highlights is another story..

I think that the "range" analysis gives you something that is not easily available at the moment; you can see how far the shadows are and might adjust the exposure based on this.

As I wrote in the first post this is currently purely prototype and an idea that I want to test in real life to see if it is up to anything. Soon others can test as I publish this for beta testing.

 

[koraks]

An area for improvement and to address the issue of high brightness scenes is to actually make two exposures, e.g. one at +2 and one at -2 stops and then combine the data. Take highlight readings from the -2 exposure and shadow readings from the +2 exposure. I'm not sure if the camera API allows your app to apply an exposure adjustment, but the camera app in my android phone does have a pro mode that allows the exposure to be compensated manually, so I suspect this might be possible on iPhones as well.
You'll likely get some minor overlay issues in low light levels if the camera/phone isn't held still, but it shouldn't be much of an issue as you don't really need high resolution for this anyway.
You could of course start out with a single exposure and enter dual exposure mode only if the SBR exceeds a certain number of stops.
Btw I also second the suggestion of changing the colors into something more intuitive. Eg red vs blue or yellow vs blue might work well. If I were to do this I'd do a quick read up on different types of color blindness in people and choose colors that they're most likely to be able to discern as distinct hues.

 

[radiant]

An area for improvement and to address the issue of high brightness scenes is to actually make two exposures, e.g. one at +2 and one at -2 stops and then combine the data.

Yes, that is exatcly what I was talking before. I can control the iPhone camera, setting exposure time etc. So it is totally doable. Now it is on auto for easyness.

And maybe later on auto-switch to dual-exposure if needed is a good idea. Lets see if it is needed first!

Btw I also second the suggestion of changing the colors into something more intuitive.

Hah, yes of course. The current colors are just "lets put something". I was thinking of following ideas:

- Visualize zones; either with their correct shades of gray or with some other indicator.
- Indicate zone V clearly
- Indicate zones I-II in "hashed red" as well as the two brightest zones IX and X

I know, I know. f-stops cannot be directly converted to zones, or so I heard. But hold on; what I'm meaning is to areas which would be either destroyed or compressed if using suggested exposure - helping to understand how exposure is going to work. For example -4 stops and below and +4 and above? Maybe this should be configurable; one could modify the values for slide film.

 

[koraks]

Ah yes, sorry, I missed the essence of your earlier ideas. Sounds good!

Yes, I know what you mean by the fstop vs zone issue. Maybe it would be nice to be able to set the width of each zone in terms of fstops so that those who do expansion & contraction can taylor the output to their needs. By default I'd probably use something like anything below -2 stops will be considered lost and anything above something like 5 stops. But this is a slippery slope and rather personal.

 

[radiant]

By default I'd probably use something like anything below -2 stops will be considered lost and anything above something like 5 stops. But this is a slippery slope and rather personal.

What puzzles me at the moment is that typically the brighest "block" is 1-2 stops higher than what I get when exposure is calculated from the whole picture. The units in this comparison should be in par; just a smaller block and larger block.

Maybe it has something to do with the cameras dynamic range; the camera cannot register higher EV readings on the parts that should be brighter..

edit: which makes me think if this can be realiable at all without some HDR trickery..

 

[grain elevator]

Great idea!
Could you include a function to change the size of the rectangles in which brightness is averaged? Perhaps a function to exclude selected rectangles from calculations? I also wonder how you average within the rectangles. Simple mean? If so, is there a provision to exclude outliers?
How about this for display: the app calculates an exposure based on whatever preferences on has selected. Then, I would like to see the little rectangles not tinted in zone values or such, as I suspect that would be harder to view, especially in sunlight, than visualising it in one's mind. Rather I'd like to see "+X" or "-X", X being the distance in stops from middle gray (or perhaps the chosen exposure index such as IRE), with just those outside of the (pre-selected) recordable range colored as a warning.

If you expose based on measurement at the "dimmest" block that area becomes "correctly" exposed. However what happens to your highlights is another story..

I believe most people mean something like placing it in zone II or III or on IRE index, not "correct" (Zone V) exposure when they say "basing exposure on shadows". In a similar vein:

What puzzles me at the moment is that typically the brighest "block" is 1-2 stops higher than what I get when exposure is calculated from the whole picture.

Isn't ~2 stops higher about right? After all you want the brightest block to be rendered brighter than middle gray?
The terminology issues make this stuff so much harder!

 

[Bill Burk]

Tap to ignore a reading, for example if you want to rule out the sky from the calculation because you are ok with it blocking

 

[koraks]

Rather I'd like to see "+X" or "-X"

That would be useful IMO for user-selectable "spots". Tap a spot and you get a little box that says + or - so many stops compared to the selected base exposure, which could be based on also a user-selectable spot. But I think I'm now basically projecting my multi-spot Canon T90 workflow onto this...then again, it's my favorite metering system and the main reason why I mostly lug the T90 around when shooting 35mm...

Tap to ignore a reading, for example if you want to rule out the sky from the calculation because you are ok with it blocking

Yes, that's very useful too I think.

 

[radiant]

Could you include a function to change the size of the rectangles in which brightness is averaged? Perhaps a function to exclude selected rectangles from calculations?

Done! Selectin/deselecting area is not done yet but it is a good idea!

also wonder how you average within the rectangles. Simple mean? If so, is there a provision to exclude outliers?

It is just average of all pixels. Same goes to "overall" exposure. Define outliners

I implemented new color coding; only groups that are over/under the configurable limit are shown. This method is really promising, quickly tried in indoor lightning and it brings attention to dark/bright areas for sure! How does it work in higher dynamic scenes, we will see tomorrow. Thankfully there are many clear sky+sunny+snow days ahead, good for testing (and not so good for photography).

Isn't ~2 stops higher about right? After all you want the brightest block to be rendered brighter than middle gray?

That would be only zone VII? Isn't correct exposure actually the exposure for zone V?

Tap to ignore a reading, for example if you want to rule out the sky from the calculation because you are ok with it blocking

Thanks for the idea! It is really interesting.

That would be useful IMO for user-selectable "spots". Tap a spot and you get a little box that says + or - so many stops compared to the selected base exposure, which could be based on also a user-selectable spot.

As the analysis is live, there is so much going on the screen, a little movement on camera changes exposure and some "border line" groups might change color back and forth. I tried color coding for configurable highlights/shadows and it might be enough.

I'm now basically projecting my multi-spot Canon T90 workflow onto this...then again, it's my favorite metering system and the main reason why I mostly lug the T90 around when shooting 35mm...

Minolta Dynax 800si here for same purposes. The metering on these "modern" SLRs is so good.

 

[Lachlan Young]

It is "correct" expsure for the highlights. I believe if that area was clear sky,it would be exposed to zone V.

When looking at the values of the app, remember all values are exposure suggestions for the selected group of pixels. There is also a "expose at" which calculates all the pixels together - averaging the scene.

I didn't quite understand the idea or my app is doing that at the moment. As it shows the "dimmest" group of pixels in terms of correct exposure, isn't this what is wanted or can you elaborate this in perspective what the app currently does? This is actually why I posted here because I'm not that advanced in exposure measurement.

I've attached some very quick phone pics of how highlight and shadow keying works using the IRE scale - they're very cropped phone pictures done under bad lighting, so apologies for the focus being questionable (I also rapidly realised that you need three hands to demonstrate it properly and take the picture...), but they show the important details... Bear in mind too that the meter I used for these pictures acquired a homemade (and very well stuck down) zone sticker at some point in its past, which I've cut away to expose the IRE 10 ('1') and 100 ('10') engravings. 'Highlight indexed' shows how you would index a detailed highlight that has an EV of 15 to IRE 100 (where the scriber point is) - this is what you'd do for a narrow latitude material. 'Shadow indexed' shows how you would index a detailed shadow reading of EV 8 to IRE 10 (where the scriber point is) - this is the one that matters for negative (especially BW negative) exposure. If these images are useful to people, I'll redo them properly with a tripod & studio lighting.

 

[radiant]

I've attached some very quick phone pics of how highlight and shadow keying works using the IRE scale

Thanks for the photos, which are great. I just don't know how that thing is used and how to transform that to an app? Choices are:

- Separate ZS display which shows some information (measured range?). I bit doubt if this is useful.
- Showing zones in top of blocks (as Koraks and grain elevator discussed). Might be difficult to follow and read because of constant update on screen. Color coding works best, it is fast to read.

 

[radiant]

I've uploaded first test version and it is currently being reviewed by Apple. If it goes through, testers should get some email about testing.. More info about testflight here: https://developer.apple.com/testflight/

Just an update of the UI to others as well. Here you can see over/under exposure visualization, I've set the limit to 1 for demonstration in this case.

 

[radiant]

Now I implemented exposure lock, the app starts to take its wings. This might have some potential, really.

I'm super keen to hear betatesters experiences & thoughts!

 

[relistan]

Very interesting! @vedostuu I'm PM'ing you now, happy to test and provide feedback.

 

[Craig75]

A quick and dirty way to get rid of readings for the sun, glare, lightbulbs, and specular highlights would seem to be

If f stop range = 10.9 (assumed max dynamic range of sensor) then sun / glare / specular highlights exist in shot so delete max highlight readings from calculation and recalculate.

Or even quicker just automatically disregard any reading over 10 stops in the calculation of ev

Later you can expand dynamic range of the meter with hdr.

 

[radiant]

A quick and dirty way to get rid of readings for the sun, glare, lightbulbs, and specular highlights would seem to be

If f stop range = 10.9 (assumed max dynamic range of sensor) then sun / glare / specular highlights exist in shot so delete max highlight readings from calculation and recalculate.

Or even quicker just automatically disregard any reading over 10 stops in the calculation of ev

Later you can expand dynamic range of the meter with hdr.

Good idea. You mean the "average" for complete photo would disregard these "blocks"?

Betatesting has gone live now and I found out that there is a public link option too, so here is link if anyone wants to try out: https://testflight.apple.com/join/MfSnZoX7 - you need TestFlight app for this too.

 

[radiant]

In this first beta there might be some hickups in the start, tap on the "settings" icon (wheel) and go back and everything should be running smoothly then.. I hope.

 

[Craig75]

Good idea. You mean the "average" for complete photo would disregard these "blocks"?

Betatesting has gone live now and I found out that there is a public link option too, so here is link if anyone wants to try out: https://testflight.apple.com/join/MfSnZoX7 - you need TestFlight app for this too.

exactly.

 

[radiant]

There is a bug which shows everything in red at first, thanks @relistan for reporting.

Use the lock icon once (lock&unlock) and it should start working. I try to hunt the bug.

 

[Paul Ozzello]

I'm using ios13.7 on my iPhone 8 plus and there is no blue setting wheel on the upper left of my screen. What is the green dot on top?

I've uploaded first test version and it is currently being reviewed by Apple. If it goes through, testers should get some email about testing.. More info about testflight here: https://developer.apple.com/testflight/

Just an update of the UI to others as well. Here you can see over/under exposure visualization, I've set the limit to 1 for demonstration in this case.


View attachment 267812

 

[radiant]

I'm using ios13.7 on my iPhone 8 plus and there is no blue setting wheel on the upper left of my screen. What is the green dot on top?

I'm working on that missing wheel, I try to uploaded new build currently.

The green dot is probably the iPhones "camera on" light on status bar.

 

[radiant]

Ok, new build available. Hope it fixes the issue. Anyone want to join testing, here is the link: https://testflight.apple.com/join/MfSnZoX7