Loading...
> >

exposure strange behaviour

PadonePadone Posts: 3,786
in Daz Studio Discussion

Since I'm not the photographer kind, usually I set the exposure directly to fit the camera to my needs. But today I played around with the other controls and I noticed something odd.

1) The first picture shows what happens if I set the exposure to 14. It changes the shutter speed to 256 and renders a darker picture than the default.

2) The second picture shows what happens if I set the ISO to 200. It changes the exposure to 14 and renders a brighter picture than the default.

Now, since the two pictures have the same exposure, that's 14, shouldn't they be the same ? Is this a bug or is there something I'm missing ?

exposure-14.jpg
644 x 206 - 36K
iso-200.jpg
641 x 204 - 38K

Comments

  • Richard HaseltineRichard Haseltine Posts: 102,436

    Bug report please - that is, open a Techncial Support ticket.

  • FishtalesFishtales Posts: 6,162

    That is normal.

    Setting the Exposure Value to 14 in the first example has increased the shutter speed so less light is reaching the render making the render dark.

    Setting the ISO to 200 hundred in the second example is allowing more light in so setting the exposure value to 14 brightens the image.

    I don't use the Exposure Value because it isn't as precise as using the individual sliders it is an overall setting.

  • agent unawaresagent unawares Posts: 3,513

    No, this is a bug, not normal. Exposure value is not responding to the ISO settings correctly. It should have gone down to 12 with the ISO change, not up to 14.

  • PadonePadone Posts: 3,786
    edited October 2017

    @Richard

    I'll do it, thank you.

    @Fishtales "That is normal"

    I do not agree at all. And I know you are very good at photography so I can't get how you miss this. The exposure triangle iso + shutter speed + f-stop has the only purpose of setting the exposure value. It doesn't matter how you get it. Two pictures having the same exposure do have to look the same. Then they may vary in depth of field and motion blur, and/or film graininess. That's what the settings are for in a real camera.

    @agent unawares

    I may be wrong but I feel it's rather the opposite. I say so because when you increase the exposure you should get a brighter image in my opinion, not a darker one. My guess is that when you set the exposure directly you get the opposite of what you do. Increasing to +1 you get -1. While setting the iso seems to work fine since iso 200 is brighter than iso 100 as it should. For sure there's something odd.

    Post edited by Padone on
  • FishtalesFishtales Posts: 6,162

    In Iray the Exposure Value is only based on the Shutter Speed and F/stop at an ISO of 100 so changing the ISO changes how that formulae is used.

    Try it. Only change the ISO and reset the Exposure Value to 14 it will change the Shutter speed to match the F/stop. Now reset everything and change the F/stop and then set the Exposure to 14 and again it will reset the Shutter speed to match the F/stop.

    https://en.wikipedia.org/wiki/Exposure_value

  • PadonePadone Posts: 3,786
    edited October 2017

    @Fishtales "only change the ISO and reset the exposure value"

    I tried it, it doesn't work at all. Below the two images. The first is everything at default with exposure 13. The second is with ISO 200 and the exposure set back to 13. It adjusts the shutter speed as you say. But it does it wrong.

    Since I doubled the ISO, to match the exposure it should double the shutter speed from 1/128 to 1/256. Instead it halves it from 1/128 to 1/64, so the image gets brighter.

    If I set the shutter speed by hand to 1/256 as it should be, then it changes the exposure to 15. But the rendered image looks at exposure 13 the same as the default. Look at the third picture. So the indicated exposure is wrong.

     

    UPDATE: I also filed the ticket.

    iso-100.jpg
    633 x 207 - 31K
    iso-200.jpg
    633 x 208 - 31K
    iso-200-bis.jpg
    628 x 207 - 32K
    Post edited by Padone on
  • FishtalesFishtales Posts: 6,162

    That's right. You double the ISO it halves the shutter speed because the formula is based on an ISO of 100 :) You increased the shutter speed which reduced the light entering. If you change the ISO to 400 and set the Exposure to 14 it quarters the Shutter Speed to 64 as it should. When using the Exposure you should leave the ISO at 100 for the formula to work correctly, that is one reason I don't use it to set the exposure for a render.

    Setting the defaults and changing the Exposure to 14 changes the speed to 256 which is wrong as it should be 250 according to the formula on the Wikipedia page.

  • PadonePadone Posts: 3,786
    edited October 2017

    @Fishtales

    I assume the formula you refer to is EV = log2(N^2/t) where EV is the exposure, N is the f-number and t is the exposure time in seconds.

    Now, doubling the film ISO is the same as doubling the aperture. And doubling the aperture is the same as halving N^2 in that formula. Because the f-number N gets smaller as the aperture increases https://en.wikipedia.org/wiki/F-number. So to maintain EV constant the exposure time t must be halved too. And to halve t you have to double the shutter speed.

    In my opinion what confuses you, is that N is not the aperture, it is the f-number. And t is not the shutter speed, it is the exposure time. Indeed DAZ Studio does the same error as you do. Apart from numbers anyway, it is obvious that DS is wrong because two pictures with the same exposure do have to look the same.


    This is the formula for the default values in DS.

    13 = log2(8^2/(1/128))


    And this is the formula when you double the ISO, or the aperture, so N goes from 8 to 5.6. As you see yourself to keep the exposure constant the exposure time must be 1/256, not 1/64.

    13 = log2(5.6^2/(1/256))
     

    Post edited by Padone on
  • FishtalesFishtales Posts: 6,162

    With a film ISO of 200 and an F/stop of 5.6 setting the EV at 14 gives a shutter speed of 261.22 in DAZ Studio instead of 256. I just tried it :) The formula in Studio is slightly off as I said.

  • ToborTobor Posts: 2,300
    edited October 2017
    Padone said:

    UPDATE: I also filed the ticket.

    You've wasted your time, as I did when I reported this back in 4.8. Either Daz thinks the behavior is correct, or it's a value they are getting back from Iray, and Iray gets it wrong and therefore there isn't anything they can do about it. In the end they simply dropped the ticket. If you look back in the threads you'll see this issue has been reported a number of times.

    Firshtales, re-adjusting values to then match the expectation of the formula sort of misses the point if you are assuming this will work like an automatic DSLR.

    In any case, it's a useless exercise because adjusting ISO doesn't do anything relevent anyway. It's not like a higher ISO will produce more film grain (in a DSLR it would increase noise, as it makes the sensor more sensitive to light). Save your efforts and just ignore the ISO setting. EV is for automatic cameras and does nothing of real benefits for renders, as none of these settings impart a physical change in the render physics (no DOF, no motion blur, no grain/noise). Simply keep at ISO 100 and adjust the f/stop and shutter speed to your liking. Manual reciprocity of these settings works correctly, as has been amply demonstrated over the years.

    Post edited by Tobor on
  • PadonePadone Posts: 3,786

    @Fishtales

    You're right, in this last example you did it works fine. It also works fine if you adjust only the ISO leaving the other parameters alone. But there are other cases, as in my examples above, where it doesn't work. It mainly breaks when you adjust the exposure directly.

    @Tobor

    I didn't know it was reported before I just followed what Richard said. I hope they'll fix it anyway, and if not, well I guess we can go on without just fine ..

  • FishtalesFishtales Posts: 6,162

    @Tobor

    I don't :)

    I use the settings for my eye not the formula. I also notice subtle differences in how the light changes by using the different options. The one time I try real world settings is when I use one of my photographs as a backdrop. By setting them to the same as in the Exif data, setting the lat/long to where the image was taken, time of day, making the camera height to roughly 2 metres and pointing it in the general compass direction I get a pretty good replication of the original lighting. It still needs a bit of manipulation of the camera to get the shadows right though :) I find it is still a good starting point.

  • ToborTobor Posts: 2,300
    edited October 2017

    Some simple experiments to show how this should work:

    EV's get smaller with greater (more bright) exposure. If ISO=100, f=8, and speed=1/125, EV is correctly indicated as 13. D|S is correct to this point, but not if you fiddle with ISO.

    * Change the aperture from f/8 to f/5.6 (doubling the brightness), which produces an EV of 12, which is correct, and also matches the wiki page. (Reset it back to f/8 before proceeding.)

    * Change ISO to 200. This also has the effect of doubling the brightness because the film/sensor is now twice as sensitive, it collects double the light. D|S INCORRECTLY shows EV=14. A higher EV means less exposure, but how could it when you've doubled the exposure sensitivity of the film?

    ISO, aperature, and shutter speed are reciprocal; adjusting one up while adjusting either of the other two down should result in the exact same exposure. This is simple physics. For a given exposure, if you make the film twice as sensitive, you have to stop down either aperture or shutter. 

    An interactive online calculator to check your results:
    https://rechneronline.de/exposure/

    The following two settings *should* give you the exact same exposure, because they are reciprocal:

    ISO=100
    f=8
    shutter=1/125
    Calculator EV=13 (correct); DS EV=13 (correct)

    ISO=200
    f=11 (reciprocal of ISO 100 to 200)
    shutter=1/125
    Calculator EV=12.9 (correct; close enough to 13); DS EV=14.92 (incorrect)

    The problem may be that some resources peg ISO at 100 when using a printed chart for calculating EV, even when using a faster or slower film/sensor. This doesn't mean ISO is not considered in calculating the exposure; it's merely another way of presenting the numbers in chart, and it expects the user will manually translate the other reciprocal settings accordingly.

    https://www.scantips.com/lights/evchart.html

    As the calculator there says (for the default settings at ISO=100): "ISO matches settings to the scene, and the Equivalent Exposures at ISO 200 are in the EV chart row at EV 14." This is for referring to a chart where EV=100. I doubt this is what users are expecting if they're used to how their digital cameras work. They aren't looking at charts, but expecting their camera to be automatically doing all the right calculations.

    I haven't messed with this stuff since I first discovered this behavior in 4.8. I have no idea if/how the problem persists in the latest betas. Since the "autocalculated" tone mapping values are not valid when usoing ISO -- even if the exposure might be -- I've simply taken to using only aperture and shutter speed. 

    Post edited by Tobor on
  • agent unawaresagent unawares Posts: 3,513

    Padone said:

    @Richard

    I'll do it, thank you.

    @Fishtales "That is normal"

    I do not agree at all. And I know you are very good at photography so I can't get how you miss this. The exposure triangle iso + shutter speed + f-stop has the only purpose of setting the exposure value. It doesn't matter how you get it. Two pictures having the same exposure do have to look the same. Then they may vary in depth of field and motion blur, and/or film graininess. That's what the settings are for in a real camera.

    @agent unawares

    I may be wrong but I feel it's rather the opposite. I say so because when you increase the exposure you should get a brighter image in my opinion, not a darker one. My guess is that when you set the exposure directly you get the opposite of what you do. Increasing to +1 you get -1. While setting the iso seems to work fine since iso 200 is brighter than iso 100 as it should. For sure there's something odd.

    Nope, because EV is not a direct measure of exposure time. Very dark scenes you shoot with a very low EV (into the negatives even), and very bright scenes you shoot with a very high EV.

    https://en.wikipedia.org/wiki/Exposure_value

  • TheKDTheKD Posts: 2,696

    I personally stopped bothering with ISO and EV, and adjust only fstop for brightness. Saves me headaches of having to memorize so much stuff.

  • PadonePadone Posts: 3,786
    edited October 2017

    @Tobor

    Thank you very much for the information you posted. As I said I'm not the photographer kind I just played around a little with this. So it's all very interesting to me. Also may be I understood a litte bit how it works.

    1) At default values everything is fine, that is, all the values are right by the formula. See picture 3.

    2) Setting the exposure to 14 the shutter speed goes to 1/256, that's right by the formula. The picture is also rendered right since it's darker than the default as it should be. See picture-4.

    3) Setting iso to 200 the exposure goes to 14, that's wrong because it should be 12 by the formula. Nevertheless the picture is rendered indeed at exposure 12. You can say that because it's brighter than the default as it should be. Further proof is that if you set the exposure manually to 12 you get the same picture. So the number is wrong but the picture is right. See picture-5.

    That stills odd but at least I can understand what's going on, and what's wrong and right. So thank you again.


    @agent unawares

    Yes now I understand better, you're right. Setting the exposure manually works fine. Setting the iso also works fine but it just displays the wrong numbers.

     

    EDIT: in the tests I used the calculator pointed by Tobor https://rechneronline.de/exposure/ that takes into account the ISO variations.

    picture-3.jpg
    517 x 205 - 34K
    picture-4.jpg
    513 x 206 - 33K
    picture-5.jpg
    519 x 207 - 35K
    Post edited by Padone on
  • ToborTobor Posts: 2,300

    Wish we could pin your message, Padone, because it clearly and concisely shows what's happening. If ISO is involved in EV (which it has to be, because it changes when you modify ISO), an EV value of 14 should give you the same exposure regardless of the settings you used to get there. 

    As I noted in my earlier post, since I noticed this in 4.8 I've taken to simply ignoring the ISO and EV values, and use only aperature and shutter controls. From my early film days,  I didn't get the chance to change the ISO mid-roll (ISO was then called ASA, showing my decrepit age), so everything I did used only f-stops and shutter speeds. The "limitation" of using just these two settings comes naturally to me anyway.

  • FishtalesFishtales Posts: 6,162

    As I said above, changing the ISO breaks the formulae because it is set for ISO 100.

    This is from the bottom of the Wikipedia page. Try the last example. Set the EV to 11 at 400 then change the 400 to 100 and it changes to EV 9 as it says it should do in the formula. I have never used the EV setting as I prefer to use the other settings to get the light right.

    "The exposure values in Table 2 are for ISO 100 speed ("EV100"). For a different ISO speed S, increase the exposure values (decrease the exposures) by the number of exposure steps by which that speed is greater than ISO 100, formally

    {\mathrm {EV}}_{{S}}={\mathrm {EV}}_{{100}}+\log _{2}{\frac {S}{100}}\,.

    For example, ISO 400 speed is two steps greater than ISO 100:

    {\mathrm {EV}}_{{400}}={\mathrm {EV}}_{{100}}+\log _{2}{\frac {400}{100}}={\mathrm {EV}}_{{100}}+2\,.

    To photograph outdoor night sports with an ISO 400–speed imaging medium, search Table 2 for "Night sports" (which has an EV of 9 for ISO 100), and add 2 to get EV400 = 11.

    For lower ISO speed, decrease the exposure values (increase the exposures) by the number of exposure steps by which the speed is less than ISO 100. For example, ISO 50 speed is one step less than ISO 100:

    {\mathrm {EV}}_{{50}}={\mathrm {EV}}_{{100}}+\log _{2}{\frac {50}{100}}={\mathrm {EV}}_{{100}}-1\,.

    To photograph a rainbow against a cloudy sky with an ISO 50–speed imaging medium, search Table 2 for "Rainbows-Cloudy sky background" (which has an EV of 14), and subtract 1 to get EV50 = 13.

    The equation for correcting for ISO speed can also be solved for EV100:

    {\mathrm {EV}}_{{100}}={\mathrm {EV}}_{{S}}-\log _{2}{\frac {S}{100}}\,.

    For example, using ISO 400 film and setting the camera for EV 11 allows shooting night sports at a light level of EV100 = 9, in agreement with the example done the other way around above. An online calculator that implemented this calculation was available at dpreview.com"

  • PadonePadone Posts: 3,786
    edited October 2017

    @Fishtales

    Yes I did take that into consideration in my last test. I used the calculator pointed by Tobor, that also includes the ISO. That calculator uses the formula below.

    EV = log2(100*N^2/(ISO*t))

    I believe the formula at wikipedia has a typo error. Since it wouldn't compensate for the ISO otherwise. So the plus should be minus. I know the two formulas don't look the same but they are. Since log sums can be transformed into multiplications.

    EV = log2(N^2/t)-log2(ISO/100)

    12 = log2(100*8^2/(200*(1/128)))
    12 = log2(8^2/(1/128))-log2(200/100)

    https://www.dpreview.com/forums/post/9729366


    Proof that the formula at wikipedia is wrong, is that by definition of the exposure triangle, doubling the aperture (from f/8 to f/5.6) must be the same as doubling the ISO (from 100 to 200). But if you use the formula at wikipedia then it's not.

    14 = log2(8^2/(1/128))+log2(200/100)
    12 = log2(5.6^2/(1/128))+log2(100/100)

    While if you use the correct formula then it's the same as it should be.

    12 = log2(8^2/(1/128))-log2(200/100)
    12 = log2(5.6^2/(1/128))-log2(100/100)

    https://fstoppers.com/education/exposure-triangle-understanding-how-aperture-shutter-speed-and-iso-work-together-72878

    Post edited by Padone on
  • agent unawaresagent unawares Posts: 3,513
    Fishtales said:

    As I said above, changing the ISO breaks the formulae because it is set for ISO 100.

    Huh. Apparently ISO kind of normalizes EV by moving the zero point. That's really counterintuitive.

  • PadonePadone Posts: 3,786
    edited October 2017

    @agent unawares

    No it's not. The formula at wikipedia is wrong. If you apply it to different ISOs it breaks. See my post above for the proof. I don't talk about the first formula at ISO 100. I talk about the second formula that should take ISO into account, but it doesn't.

    EVs = EV100+log2(S/100)

    it has to be corrected into

    EVs = EV100-log2(S/100)

    This is the same formula used in the calculator pointed by Tobor, and it works fine. That is, it doesn't break when you change the ISO. That is, if we can agree than in the exposure triangle doubling the ISO must be the same as doubling the aperture (or halving the shutter speed), then the formula at wikipedia is wrong.

    Post edited by Padone on
  • ToborTobor Posts: 2,300
    edited October 2017

    The traditional formula for EV doesn't take into consideration ISO. It's either left out and assumes users are interpreting the results using charts or manual additional calculation -- a dubious assumption with any automatic system -- or ISO is considered as part of the output, and integrated as such. It becomes a matter of expectation. Since numerous users have reported the D|S behavior as incorrect, there's a strong argument that displaying EVs based on ISO=100 is the wrong approach.

    If you know your EVs D|S will either confuse you or make you do additional work you shouldn't have to do with an automatic system. It's as simple as that. We've already demonstrated that the actual exposure is correct, even though the EV value displayed is based on ISO=100, regardless of the ISO setting.

    We're all referencing the same Wikipedia page and online calculators. They are all correct in their own way, but we're talking in cross-purposes. My argument, stated previously, is that there is no purpose in constraining EV to ISO=100 as a calculation in an automated system. Standardizing at ISO=100 was meant for printed charts. Why would we be using prepared charts to set Iray exposure? That seems silly to me.

     

    Post edited by Tobor on
  • PadonePadone Posts: 3,786

    UPDATE: Here is the reply of the support team just to be useful to anyone interested. I do not agree and I told them but I can't do anything more since they closed the ticket. Anyway this has been a very interesting and useful discussion. At least to me. So thank you all.

     

    Emma H Today at 17:33

    Hi Alessandro,

    I went and spoke to our Dev's this morning. They have seen that forum thread and have looked into this. That Exposure Value number is not what Iray uses when rendering. Iray uses the 3 other values (Shutter Speed, F/Stop, and Film ISO.) Even if it is wrong, it is not affecting your renders. You will want to ignore the Exposure Value as it is just a shortcut to help people. Below is the formula our Dev's use to calculate this. Please note that they did test this and it is correct.

    }image
    https://en.wikipedia.org/wiki/Exposure_value

    Thank you,
    Emmalee

  • agent unawaresagent unawares Posts: 3,513
    Padone said:

    @agent unawares

    No it's not. The formula at wikipedia is wrong.

    Ehhh I didn't come to that conclusion from Wikipedia or Fishtales but from photography sites that also state doubling the ISO raises the EV by one. No, this doesn't make any sense to me either. It's absurdly unintuitive. Maybe they are wrong.

  • PadonePadone Posts: 3,786
    edited October 2017

    Just for completeness.

    And if you check it, EVs = EV100 - log2(S/100) is indeed the formula iray uses to render the picture. As in my example above in picture 5 where the image is at exposure 12 (by iray) while the render panel says exposure 14 (by wikipedia).

    This is more proof that the correct formula is EVs = EV100 - log2(S/100). Because this formula matches what iray does in the picture. You can try with any parameter it always matches.

    Post edited by Padone on
Sign In or Register to comment.