On Tue, May 24, 2016 at 10:01 AM, johannes hanika <hana...@gmail.com> wrote:
> On Wed, May 25, 2016 at 4:23 AM, J. Liles <malnour...@gmail.com> wrote: > > > > > > On Tue, May 24, 2016 at 9:18 AM, Roman Lebedev <lebedev...@gmail.com> > wrote: > >> > >> On Tue, May 24, 2016 at 6:44 PM, J. Liles <malnour...@gmail.com> wrote: > >> > > >> > > >> > On Tue, May 24, 2016 at 5:40 AM, johannes hanika <hana...@gmail.com> > >> > wrote: > >> >> > >> >> hi, > >> >> > >> >> the good news is your black levels look very uniform there, so no > >> >> change required. > >> >> > >> >> if you have a sample where these are actually different.. rawspeed > >> >> only supports a hardcoded amount of 4 black levels today. no big deal > >> >> changing it, but without a use case i don't think we should. > >> >> > >> >> cheers, > >> >> jo > >> >> > >> >> > >> > > >> > In every one I've looked at, all the values have matched. At least > with > >> > the > >> > X70 and the shooting modes I've tried. > >> > > >> > Indeed, locally I changed the == 4 to >= 4 when interpreting this tag > in > >> > rawspeed's RafDecoder.cpp and that seemed to do the trick to get the > >> > first 4 > >> > black levels into darktable. > >> Hm, and without this change, what values does the darktable receive? > >> Something very different? > > > > > > Without this, I believe it uses default values specified in cameras.xml. > > > > I think it's always using 1024, whereas the RAF files specify a different > > black level for each ISO (IIRC between 1023 and 1026). > > uh, so that sounds like it does make some sense to try and support it. > i guess the clean way would be to put all 36 (!) values into > rawprepare params :(. without proof that it's ever needed i'd rather > go with precise black levels in cameras.xml (these can depend on iso, > too), or at least use the first black level as a constant for all > colour channels. > > are these values strictly only dependent on iso? > > -jo > > They do not appear to just depend on ISO (maybe exposure time is part of it too?) I have several ISO 1600 images with different black level values. It also depends on camera/sensor. Ingo's problematic tree image is from the X-Pro1 and has the same 36 value slots for black level, but the ISO 200 level is 255 instead of 1023. I checked 1188 RAF files and every one of them has identical values for all 36 black levels. I think it's reasonable to just use the first value. In order to map the first four correctly you'd need to use the X-Trans Layout tag to figure out which entries correspond to R G G and B anyway. So I think the current solution for the X100 is probably technically wrong. But it seems very likely that the values are all the same anyway so it doesn't matter (yet?) The RAF files from X-Trans III cameras that I've looked at appear to follow the same pattern as the X-Trans II files. > > > > >> > >> > >> >> > >> >> On Mon, May 23, 2016 at 8:36 PM, J. Liles <malnour...@gmail.com> > wrote: > >> >> > FYI this change does not help with the Fuji X70 black levels. > Here's > >> >> > what > >> >> > the black level tag looks like for the X70 and other newer fuji > >> >> > cameras: > >> >> > > >> >> > Black Level : 1026 1026 1026 1026 1026 1026 > 1026 > >> >> > 1026 > >> >> > 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 > 1026 > >> >> > 1026 > >> >> > 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 1026 > >> >> > > >> >> > > >> >> > On Sun, May 15, 2016 at 10:27 AM, johannes hanika < > hana...@gmail.com> > >> >> > wrote: > >> >> >> > >> >> >> see > >> >> >> > >> >> >> https://github.com/darktable-org/darktable/pull/1202 > >> >> >> > >> >> >> j. > >> >> >> > >> >> >> On Mon, May 16, 2016 at 5:02 AM, Martijn Kruiten > >> >> >> <mart...@kruiten.nl> > >> >> >> wrote: > >> >> >> > Yes, I thought it was amazing too. Thanks for looking into it! > >> >> >> > > >> >> >> > Op zo 15 mei 2016 om 17:40 schreef johannes hanika > >> >> >> > <hana...@gmail.com>: > >> >> >> >> > >> >> >> >> oh, right. thanks for pointing that out. amazing what a > >> >> >> >> difference > >> >> >> >> this small change makes. i shall look into pushing that into > >> >> >> >> rawspeed's camera.xml then. > >> >> >> >> > >> >> >> >> -j. > >> >> >> >> > >> >> >> >> On Mon, May 16, 2016 at 3:14 AM, Martijn Kruiten > >> >> >> >> <mart...@kruiten.nl> > >> >> >> >> wrote: > >> >> >> >> > Correction: it obviously varies with ISO /and/ some other > >> >> >> >> > variables. > >> >> >> >> > > >> >> >> >> > Op zo 15 mei 2016 om 17:10 schreef Martijn Kruiten > >> >> >> >> > <mart...@kruiten.nl>: > >> >> >> >> >> > >> >> >> >> >> Your blackpoint was not the same as the value that was > >> >> >> >> >> reported > >> >> >> >> >> in > >> >> >> >> >> the > >> >> >> >> >> Google Groups issue. I changed it to be the same and that > >> >> >> >> >> fixed > >> >> >> >> >> it, > >> >> >> >> >> so > >> >> >> >> >> that's definitely where the issue is > >> >> >> >> >> > >> >> >> >> >> > >> >> >> >> >> > >> >> >> >> >> > >> >> >> >> >> ( > https://www.dropbox.com/s/d3h5uwwoj7xe4w7/DSCF3831-Darktable_blackpointfix.png?dl=0 > ). > >> >> >> >> >> > >> >> >> >> >> Quote from the group: > >> >> >> >> >> > >> >> >> >> >> "Camconst.json cannot help with BL because it is not steady > in > >> >> >> >> >> every > >> >> >> >> >> ISO > >> >> >> >> >> but varies depending on other settings (exposure time should > >> >> >> >> >> be > >> >> >> >> >> one), > >> >> >> >> >> sensors temperature etc. > >> >> >> >> >> > >> >> >> >> >> I converted the samples provided (see forum link) to DNG and > >> >> >> >> >> there > >> >> >> >> >> one > >> >> >> >> >> can > >> >> >> >> >> see that while the first 3200ISO sample (DSC3831) has BL > >> >> >> >> >> 253-258-258-253 the > >> >> >> >> >> other two samples at 3200ISO have BL 251-260-260-251, > >> >> >> >> >> 251-263-263-250 > >> >> >> >> >> (channel order is R-G1-G2-B) .." > >> >> >> >> >> > >> >> >> >> >> So the blackpoint varies with ISO and maybe some other > >> >> >> >> >> variables > >> >> >> >> >> and > >> >> >> >> >> should be read out for each image independently. > >> >> >> >> >> > >> >> >> >> >> Regards, > >> >> >> >> >> > >> >> >> >> >> Martijn > >> >> >> >> >> > >> >> >> >> >> Op zo 15 mei 2016 om 16:43 schreef johannes hanika > >> >> >> >> >> <hana...@gmail.com>: > >> >> >> >> >>> > >> >> >> >> >>> well, we're not using dcraw and the black point (as you can > >> >> >> >> >>> see > >> >> >> >> >>> in > >> >> >> >> >>> the > >> >> >> >> >>> screenshot) is the same as the one in the google groups > issue > >> >> >> >> >>> you > >> >> >> >> >>> posted after the fix. possibly the black point is a red > >> >> >> >> >>> herring > >> >> >> >> >>> here. > >> >> >> >> >>> > >> >> >> >> >>> j. > >> >> >> >> >>> > >> >> >> >> >>> On Mon, May 16, 2016 at 2:32 AM, Martijn Kruiten > >> >> >> >> >>> <mart...@kruiten.nl> > >> >> >> >> >>> wrote: > >> >> >> >> >>> > Hi Johannes, > >> >> >> >> >>> > > >> >> >> >> >>> > I only pushed a few stops to exaggerate it. I'm seeing > >> >> >> >> >>> > green > >> >> >> >> >>> > shadows > >> >> >> >> >>> > even in > >> >> >> >> >>> > well lit daylight ISO 1600 shots without exposure > >> >> >> >> >>> > compensation, > >> >> >> >> >>> > but > >> >> >> >> >>> > i > >> >> >> >> >>> > felt > >> >> >> >> >>> > this image could demonstrate it much better. Also, in > your > >> >> >> >> >>> > screenshot > >> >> >> >> >>> > the > >> >> >> >> >>> > shadows are still green, just less obvious because they > are > >> >> >> >> >>> > darker. > >> >> >> >> >>> > They are > >> >> >> >> >>> > not green in Adobe Camera Raw, DxO and Aperture and they > >> >> >> >> >>> > aren't > >> >> >> >> >>> > in > >> >> >> >> >>> > RawTherapee either, all at default settings, white > balance > >> >> >> >> >>> > as > >> >> >> >> >>> > shot, > >> >> >> >> >>> > exposure > >> >> >> >> >>> > compensation set at any value, camera profile set to any > >> >> >> >> >>> > profile. > >> >> >> >> >>> > I > >> >> >> >> >>> > added a > >> >> >> >> >>> > Camera Raw sample to the Dropbox folder as an additional > >> >> >> >> >>> > sample. > >> >> >> >> >>> > If > >> >> >> >> >>> > it's > >> >> >> >> >>> > just a limitation of the way Darktable processes these > >> >> >> >> >>> > images, > >> >> >> >> >>> > so > >> >> >> >> >>> > be > >> >> >> >> >>> > it, but > >> >> >> >> >>> > to me it seems very much like the issue RawTherapee had > in > >> >> >> >> >>> > 2013. > >> >> >> >> >>> > Sadly > >> >> >> >> >>> > I do > >> >> >> >> >>> > not have a sample of the old RawTherapee picture anymore. > >> >> >> >> >>> > I'll > >> >> >> >> >>> > check > >> >> >> >> >>> > if > >> >> >> >> >>> > I > >> >> >> >> >>> > can download an older version. It was just as bad as it > is > >> >> >> >> >>> > in > >> >> >> >> >>> > Darktable > >> >> >> >> >>> > now. > >> >> >> >> >>> > There is another (closed source) raw editor with this > >> >> >> >> >>> > issue, > >> >> >> >> >>> > but > >> >> >> >> >>> > I > >> >> >> >> >>> > don't > >> >> >> >> >>> > remember which one it was. > >> >> >> >> >>> > > >> >> >> >> >>> > Also, I'll try to set the blackpoint myself. I didn't > >> >> >> >> >>> > realize > >> >> >> >> >>> > there > >> >> >> >> >>> > is > >> >> >> >> >>> > a gui > >> >> >> >> >>> > option for that. I'll report back. Hopefully it's indeed > >> >> >> >> >>> > consistent > >> >> >> >> >>> > across > >> >> >> >> >>> > images, although I could not find the correct blackpoint > in > >> >> >> >> >>> > the > >> >> >> >> >>> > gui > >> >> >> >> >>> > of > >> >> >> >> >>> > RawTherapee before they patched DCRaw. > >> >> >> >> >>> > > >> >> >> >> >>> > Regards, > >> >> >> >> >>> > > >> >> >> >> >>> > Martijn > >> >> >> >> >>> > > >> >> >> >> >>> > Op zo 15 mei 2016 om 14:51 schreef johannes hanika > >> >> >> >> >>> > <hana...@gmail.com>: > >> >> >> >> >>> >> > >> >> >> >> >>> >> hey, > >> >> >> >> >>> >> > >> >> >> >> >>> >> i believe our black point is correct (see screenshot > below > >> >> >> >> >>> >> [0]). > >> >> >> >> >>> >> > >> >> >> >> >>> >> it's just a bit much to ask to push a 3200 ISO image by > 3 > >> >> >> >> >>> >> stops > >> >> >> >> >>> >> and > >> >> >> >> >>> >> get useful colour rendition out of the box. the > >> >> >> >> >>> >> rawtherapee > >> >> >> >> >>> >> shot > >> >> >> >> >>> >> looks > >> >> >> >> >>> >> like a different whitebalance/colour profile/curve has > >> >> >> >> >>> >> been > >> >> >> >> >>> >> applied? > >> >> >> >> >>> >> > >> >> >> >> >>> >> anyways, you're always free to change the black point > for > >> >> >> >> >>> >> such > >> >> >> >> >>> >> noisy > >> >> >> >> >>> >> shots in emergency cases. maybe if that is consistently > >> >> >> >> >>> >> the > >> >> >> >> >>> >> case > >> >> >> >> >>> >> for > >> >> >> >> >>> >> all images, we should include a high-iso black level > into > >> >> >> >> >>> >> the > >> >> >> >> >>> >> rawspeed > >> >> >> >> >>> >> cameras.xml? > >> >> >> >> >>> >> > >> >> >> >> >>> >> cheers, > >> >> >> >> >>> >> jo > >> >> >> >> >>> >> > >> >> >> >> >>> >> [0] > >> >> >> >> >>> >> > >> >> >> >> >>> >> > >> >> >> >> >>> >> > >> >> >> >> >>> >> > >> >> >> >> >>> >> > >> >> >> >> >>> >> > http://picpaste.com/pics/2016-05-16-004547_2560x1440_scrot-6zN1i4ho.1463316422.png > >> >> >> >> >>> >> > >> >> >> >> >>> >> On Mon, May 16, 2016 at 12:13 AM, Martijn Kruiten > >> >> >> >> >>> >> <mart...@kruiten.nl> > >> >> >> >> >>> >> wrote: > >> >> >> >> >>> >> > Hi, > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > There seems to be an issue with the blackpoint for > some > >> >> >> >> >>> >> > Fujifilm > >> >> >> >> >>> >> > cameras, or > >> >> >> >> >>> >> > at least the X100 model. I'm not sure if you're using > >> >> >> >> >>> >> > DCRaw > >> >> >> >> >>> >> > as > >> >> >> >> >>> >> > the > >> >> >> >> >>> >> > decoder, > >> >> >> >> >>> >> > but it had and may still have an issue that causes > this, > >> >> >> >> >>> >> > which > >> >> >> >> >>> >> > has > >> >> >> >> >>> >> > been > >> >> >> >> >>> >> > patched by RawTherapee after I reported it back in > 2013 > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > ( > https://code.google.com/archive/p/rawtherapee/issues/2049). > >> >> >> >> >>> >> > What > >> >> >> >> >>> >> > happens is > >> >> >> >> >>> >> > that the darkest parts of the image have a green tinge > >> >> >> >> >>> >> > because > >> >> >> >> >>> >> > the > >> >> >> >> >>> >> > blackpoint is read incorrectly from the raw file. This > >> >> >> >> >>> >> > is > >> >> >> >> >>> >> > unfortunate, > >> >> >> >> >>> >> > because I really like working with Darktable. I > exported > >> >> >> >> >>> >> > a > >> >> >> >> >>> >> > sample > >> >> >> >> >>> >> > with > >> >> >> >> >>> >> > the > >> >> >> >> >>> >> > exposure > >> >> >> >> >>> >> > correction set at +3 and exported the same image with > >> >> >> >> >>> >> > default > >> >> >> >> >>> >> > settings > >> >> >> >> >>> >> > from > >> >> >> >> >>> >> > RawTherapee to show what I'm talking about: > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > https://www.dropbox.com/sh/m5wtb7rfyxv1e1r/AABQ6czz7p-l9qf0Sly8093wa?dl=0. > >> >> >> >> >>> >> > The raw is also included. If I can help in any way, > >> >> >> >> >>> >> > please > >> >> >> >> >>> >> > let > >> >> >> >> >>> >> > me > >> >> >> >> >>> >> > know. > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > Regards, > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > Martijn Kruiten > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > >> >> >> >> >>> >> > > ___________________________________________________________________________ > >> >> >> >> >>> >> > darktable developer mailing list to unsubscribe send a > >> >> >> >> >>> >> > mail > >> >> >> >> >>> >> > to > >> >> >> >> >>> >> > darktable-dev+unsubscr...@lists.darktable.org > >> >> >> > >> >> >> > >> >> >> > >> >> >> > ___________________________________________________________________________ > >> >> >> darktable developer mailing list > >> >> >> to unsubscribe send a mail to > >> >> >> darktable-dev+unsubscr...@lists.darktable.org > >> >> >> > >> >> > > >> > > >> > > >> > > >> > > >> > > ___________________________________________________________________________ > >> > darktable developer mailing list to unsubscribe send a mail to > >> > darktable-dev+unsubscr...@lists.darktable.org > >> Roman. > > > > > ___________________________________________________________________________ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
