Re: [hugin-ptx] Re: tweaking enfuse parameters pays
kfj wrote: On 29 Jul., 11:25, Lukáš Jirkovský wrote: You may want to take a look at rawtherapee 3.x. It can do a very good job at restoring the highligths. I was able to get this [1] while the in camera jpeg looks like this [2] using the method described at [3]. It still has some ugly magenta tint in some places but still rt was able to restore incredible number of detail. Raw Therapee is too immature for my taste. The concept is really nice and I've tried it again and again for the past couple of years or so, but on my system it just hogs resources, takes forever and eventually crashes for no reason at some unforseen point. Odd, works flawlessly here. Furthermore I can't do [3] since my current version of Raw Therapee [3.01 alpha 1] does not have a 'RAW' tab. I use up all my masochism using and programming for hugin ;-) I'll give it a while and once it says it's something like 3.2 final I'll give it a good long try again. Am I too impatient? As far as I can see from what I get (before it crashes) blown is blown. If the sensor limit is hit there just isn't anything you can do. That is true. "Highlight recovery" works by using the extra bit depth (12) of the RAW image to try to come up with an approximate 8-bit color. The sensor in my Minolta is very quick to blow highlights, and highlight recovery doesn't work very well, if at all. -- Gnome Nomad gnomeno...@gmail.com wandering the landscape of god http://www.cafepress.com/otherend/ -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On 29 Jul., 11:25, Lukáš Jirkovský wrote: > > You may want to take a look at rawtherapee 3.x. It can do a very good > job at restoring the highligths. I was able to get this [1] while the > in camera jpeg looks like this [2] using the method described at [3]. > It still has some ugly magenta tint in some places but still rt was > able to restore incredible number of detail. Raw Therapee is too immature for my taste. The concept is really nice and I've tried it again and again for the past couple of years or so, but on my system it just hogs resources, takes forever and eventually crashes for no reason at some unforseen point. Furthermore I can't do [3] since my current version of Raw Therapee [3.01 alpha 1] does not have a 'RAW' tab. I use up all my masochism using and programming for hugin ;-) I'll give it a while and once it says it's something like 3.2 final I'll give it a good long try again. Am I too impatient? As far as I can see from what I get (before it crashes) blown is blown. If the sensor limit is hit there just isn't anything you can do. Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: tweaking enfuse parameters pays
On 28 July 2011 13:53, kfj <_...@yahoo.com> wrote: > I suppose my 450D isn't too far from a 550D, but I'll nevertheless do > a couple of test shots to see if I can also recover two stops from > what it reckons is blown. What raw converter do you use? You may want to take a look at rawtherapee 3.x. It can do a very good job at restoring the highligths. I was able to get this [1] while the in camera jpeg looks like this [2] using the method described at [3]. It still has some ugly magenta tint in some places but still rt was able to restore incredible number of detail. [1] http://blender6xx.ic.cz/pub/_MG_2054.jpg [2] http://blender6xx.ic.cz/pub/_MG_2054.thumb.jpg [3] http://rawtherapee.com/forum/viewtopic.php?t=2907 Lukas -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On Thursday, July 28, 2011 1:53:11 PM UTC+2, kfj wrote: > > On 28 Jul., 09:39, Jeffrey Martin <360c...@gmail.com> wrote: > > I suppose my 450D isn't too far from a 550D, but I'll nevertheless do > a couple of test shots to see if I can also recover two stops from > what it reckons is blown. What raw converter do you use? > > i use photoshop (adobe camera raw) -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On 29 Jul., 07:31, Yuval Levy wrote: > On July 27, 2011 05:05:41 AM kfj wrote: > > > measuring the time-to-satuartion for the cells. > > Wonder if anyone thought of that? > > http://www.google.com/patents/about?id=MkDJEBAJ&dq=measure+time+t... > > coming to a camera in your neighborhood photo shop in 17 years? Why 17? I thought what usually happens is that a novel patent is bought up by the established industry and made available by licensing terms which make sure it's never used. By the time the patent expires, it's obsolete, because some much cleverer way has been found to achieve the desired effect, but the established industry has managed to sell another few billions of the stuff they made. Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: tweaking enfuse parameters pays
On July 27, 2011 05:05:41 AM kfj wrote: > measuring the time-to-satuartion for the cells. > Wonder if anyone thought of that? http://www.google.com/patents/about?id=MkDJEBAJ&dq=measure+time+to+saturation+%2Bphotography coming to a camera in your neighborhood photo shop in 17 years? Yuv signature.asc Description: This is a digitally signed message part.
[hugin-ptx] Re: tweaking enfuse parameters pays
On 28 Jul., 09:39, Jeffrey Martin <360cit...@gmail.com> wrote: > careful of histogram on canon - it is a histogram from the jpeg, even when > shooting raw. what is overexposed is not really. so you can push it another > stop or more. on my 550D i can push it up to 2 stops more and recover > highlights that it says are totally blown. I wonder if this is the full story. When I look at the images, I use digiKam initially. It shows me blown highlights and I can recover them to an extent (I do not know the precise magic they use, but I suppose it's using the fact that one colour channel gets saturated first and you can extrapolate it's intensity looking at the other channels for that pixel) digiKam has no reason to tell me something is blown when it isn't - it processes in 16 bit (at least I believe it does). When it comes to Canon, you can tell the body to work in 'highlight tone priority' mode. Initially I used that, but I found out that all it does is artificially underexpose your images by one stop or so, so you can 'recover' later. I now prefer to expose on the conservative side and so protect myself against blown highlights. I suppose my 450D isn't too far from a 550D, but I'll nevertheless do a couple of test shots to see if I can also recover two stops from what it reckons is blown. What raw converter do you use? > > In fact I think the technology to expose the > > sensor for a fixed period of time and then count the photons is silly. > > ... > Heh, I wondered this 6 years ago ;-) > http://www.panotools.org/mailarchive/msg/34821#msg34821 So did I, I just never told anyone... > maybe such ideas are easy when you're not building the sensors? ;-)) I don't > know! It sure seems like a perfect idea :) We can still dream. And sometimes progress in technology comes from outsiders or fringe people because they have more freedom to think off the beaten track and sometimes have knowledge from other fields that the specialists they are 'infringing upon' lack. > BTW I had a chance to use a Leaf back on a mamiya 645DF. 16 bits of color > per channel. amazing! not your casual pocket camera though :) Lucky you. But I'm confident it won't be long until even your mobile phone will do 32 bit floating point HDR shots... I started wondering about the possibilities of stitching individual images into a larger pictuer in the late 80s, working on 512X512 grayscale frames from a frame grabber attached to a video microscope. I had a 286 processor (and, wow, a 287 coprocessor). Things took a long while... such a long while that eventually I gave up on video microscopy and trying to make sense of the images and turned to other topics. Many years later I got my first digital camera, and it could do 2048X1536 in 8 bit RGB! And I found autostitch somewhere. Was I happy. Time taught me to rather than try and force something in a bleeding edge technology waiting a few years often does the trick ;) Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: tweaking enfuse parameters pays
Jeffrey Martin wrote: On Wednesday, July 27, 2011 11:05:41 AM UTC+2, kfj wrote: On 27 Jul., 09:28, Jeffrey Martin <360c...@gmail.com> wrote: > Personally i've had the best results using +1.5 and -0.5 exposure derived > from a single raw file. Differently processed single raw files are perfect source material - if the scene can be captured with the dynamic range available. I don't know what you have, but my sensor's dynamic range is somewhere in the 12 to 14 bit range, and my landscapes sometimes just don't fit into that. I wish they would. careful of histogram on canon - it is a histogram from the jpeg, even when shooting raw. what is overexposed is not really. so you can push it another stop or more. on my 550D i can push it up to 2 stops more and recover highlights that it says are totally blown. Get to know the CCD chip in your camera. I don't know which CCD Canon uses. In my Maxxum 7D, the CCD easily blows highlights. -- Gnome Nomad gnomeno...@gmail.com wandering the landscape of god http://www.cafepress.com/otherend/ -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On Wednesday, July 27, 2011 11:05:41 AM UTC+2, kfj wrote: > > On 27 Jul., 09:28, Jeffrey Martin <360c...@gmail.com> wrote: > > > Personally i've had the best results using +1.5 and -0.5 exposure derived > > > from a single raw file. > > Differently processed single raw files are perfect source material - > if the scene can be captured with the dynamic range available. I don't > know what you have, but my sensor's dynamic range is somewhere in the > 12 to 14 bit range, and my landscapes sometimes just don't fit into > that. I wish they would. careful of histogram on canon - it is a histogram from the jpeg, even when shooting raw. what is overexposed is not really. so you can push it another stop or more. on my 550D i can push it up to 2 stops more and recover highlights that it says are totally blown. > In fact I think the technology to expose the > sensor for a fixed period of time and then count the photons is silly. > What would be much more sensible is measuring the time it takes each > cell to reach saturation. When the exposure is finally stopped, those > cells which aren't full can still be photon-counted to define the > shadows. Store the result in a floating point format and you end up > with a truly HDR raw image without any fuss and then you can proceed > by exposure-blending different versions of it instead of the > cumbersome tone-mapping. I wonder if that's technically feasible, but > why not? Heh, I wondered this 6 years ago ;-) http://www.panotools.org/mailarchive/msg/34821#msg34821 maybe such ideas are easy when you're not building the sensors? ;-)) I don't know! It sure seems like a perfect idea :) BTW I had a chance to use a Leaf back on a mamiya 645DF. 16 bits of color per channel. amazing! not your casual pocket camera though :) -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On 27 Jul., 09:28, Jeffrey Martin <360cit...@gmail.com> wrote: > Personally i've had the best results using +1.5 and -0.5 exposure derived > from a single raw file. Differently processed single raw files are perfect source material - if the scene can be captured with the dynamic range available. I don't know what you have, but my sensor's dynamic range is somewhere in the 12 to 14 bit range, and my landscapes sometimes just don't fit into that. I wish they would. In fact I think the technology to expose the sensor for a fixed period of time and then count the photons is silly. What would be much more sensible is measuring the time it takes each cell to reach saturation. When the exposure is finally stopped, those cells which aren't full can still be photon-counted to define the shadows. Store the result in a floating point format and you end up with a truly HDR raw image without any fuss and then you can proceed by exposure-blending different versions of it instead of the cumbersome tone-mapping. I wonder if that's technically feasible, but why not? Processing happens in the GHz range, that's 2 to the power of 32. Exposure times are in an order of magnitude of thousandth of seconds. So in 2 ^ -10 seconds you should have 2 ^ 22 clock cycles - that should be the dynamic range achievable just by measuring the time- to-satuartion for the cells. If the full cell triggers a store operation on the current clock value, there you go. Like a neuron firing. 2 ^ 22 seems like plenty already (can't be bothered just right now to make dB of it). Wonder if anyone thought of that? Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: tweaking enfuse parameters pays
Personally i've had the best results using +1.5 and -0.5 exposure derived from a single raw file. -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: tweaking enfuse parameters pays
kfj wrote: On 26 Jul., 10:53, Gnome Nomad wrote: kfj wrote: Hi group! Working on the images of my last trip to the Alps I spent a fair amount of time on a 360X180 of one of my favourite spots. It's a deep- cut valley with deep shadows - taken on a sunny day with a bit of fluffy cloud. No way I could get the sky and the shadows right with one exposure, so I took a set of brackets (-3,-1,+1) and made a panorama from exposure-fused stacks from it. Hmm, from my reading (and limited experience) about shooting frames to be turned into a HDR image, you shoot one frame at whatever exposure (at fixed aperture) gives proper exposure for the darkest parts, one frame at the corresponding exposure that gives proper exposure for the lightest parts, then space the intermediate exposures 2-3 stops between them. hmm to you, Gnome Nomad ;-) -3 -1 +1 is precisely that: -3 is for the lightest parts, +1 for the darkest, and the middle one (-1) ends me up with 2 stops between. What are you trying to tell me? My posting is not about shooting brackets but about processing them with enfuse. OK. I just was remembering doing some HDR sequences awhile back, and seeing ranges of 7-9 stops, and thinking that a range of 4 stops isn't very much ... As I'm on this topic again, I'd like to add another hopefully helpful hint. Even though I already put my middle exposure at -1, the +1 exposure still is way to bright, but I need it for the deep shadows. Nevertheless the result of the fusion often comes out too bright. It may just be my specific process, but I often found that on top of tweaking --exposure-sigma, lowering --exposure-mu helped - it makes the overall result slightly darker and prefers the darker exposures, but there is still an appreciable fill light from lightest exposure. So I set my default parameters for enfuse to --exposure-sigma=.1 --saturation-weight=0 --exposure-mu=.35 and often enough that hits the spot :) Cool. I still prefer to use QTPSFGui for my HDR images. I think I'd do that even if I was going to combine the resulting image into a panorama. -- Gnome Nomad gnomeno...@gmail.com wandering the landscape of god http://www.cafepress.com/otherend/ -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On 26 Jul., 15:37, kevin wrote: > I've noticed similar issues with night shots. Sometimes blown out > highlights around lights will cause those areas to be brighter then > they should and then you have to tweak some settings. I've tried > different things including using masks to combine images in gimp to > remove the blown out areas where I have other images that contain > detail, but it always seems to leave some harsh edges. Enfuse has the > --entropy-cutoff=LOWER-CUTOFF:UPPER-CUTOFF setting which will make > pixels above and below a certain point either white or black, but I > think what might be more useful would be an option where it gives > pixels above and below a certain point zero weight. If you 'steepen' the gaussian function used for weighting (by decreasing exposure-sigma), the effect approaches totally cutting off off-center values, instead of still carrying a noticable part of them into the final result. Part of the reason why I got into doing that was because the blown highlights in my brightest shot would still 'bleed through', and going to sigma values below .1, the bleedthrough would stop. Entropy isn't really what you want to use as a criterion for excluding blown highlights, but I can't seem to find a similar 'cutoff' option for the exposure, probably there isn't one because you get better (smoother) effects anyway tweaking mu and sigma. Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On Jul 26, 6:52 am, kfj <_...@yahoo.com> wrote: > As I'm on this topic again, I'd like to add another hopefully helpful > hint. Even though I already put my middle exposure at -1, the +1 > exposure still is way to bright, but I need it for the deep shadows. > Nevertheless the result of the fusion often comes out too bright. It > may just be my specific process, but I often found that on top of > tweaking --exposure-sigma, lowering --exposure-mu helped - it makes > the overall result slightly darker and prefers the darker exposures, > but there is still an appreciable fill light from lightest exposure. > So I set my default parameters for enfuse to > > --exposure-sigma=.1 --saturation-weight=0 --exposure-mu=.35 > > and often enough that hits the spot :) > > Kay I've noticed similar issues with night shots. Sometimes blown out highlights around lights will cause those areas to be brighter then they should and then you have to tweak some settings. I've tried different things including using masks to combine images in gimp to remove the blown out areas where I have other images that contain detail, but it always seems to leave some harsh edges. Enfuse has the --entropy-cutoff=LOWER-CUTOFF:UPPER-CUTOFF setting which will make pixels above and below a certain point either white or black, but I think what might be more useful would be an option where it gives pixels above and below a certain point zero weight. If all the pixels from the combined images were zero weighted then it could make that pixel white or black (depending upon what the original pixels were). This way that one image you have where you are trying to capture all the shadow detail and most of the image is blown out to white, all that white wouldn't mess up the other images that contain detail in the white areas. -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: tweaking enfuse parameters pays
On 26 Jul., 10:53, Gnome Nomad wrote: > kfj wrote: > > Hi group! > > > Working on the images of my last trip to the Alps I spent a fair > > amount of time on a 360X180 of one of my favourite spots. It's a deep- > > cut valley with deep shadows - taken on a sunny day with a bit of > > fluffy cloud. No way I could get the sky and the shadows right with > > one exposure, so I took a set of brackets (-3,-1,+1) and made a > > panorama from exposure-fused stacks from it. > > Hmm, from my reading (and limited experience) about shooting frames to > be turned into a HDR image, you shoot one frame at whatever exposure (at > fixed aperture) gives proper exposure for the darkest parts, one frame > at the corresponding exposure that gives proper exposure for the > lightest parts, then space the intermediate exposures 2-3 stops between > them. hmm to you, Gnome Nomad ;-) -3 -1 +1 is precisely that: -3 is for the lightest parts, +1 for the darkest, and the middle one (-1) ends me up with 2 stops between. What are you trying to tell me? My posting is not about shooting brackets but about processing them with enfuse. As I'm on this topic again, I'd like to add another hopefully helpful hint. Even though I already put my middle exposure at -1, the +1 exposure still is way to bright, but I need it for the deep shadows. Nevertheless the result of the fusion often comes out too bright. It may just be my specific process, but I often found that on top of tweaking --exposure-sigma, lowering --exposure-mu helped - it makes the overall result slightly darker and prefers the darker exposures, but there is still an appreciable fill light from lightest exposure. So I set my default parameters for enfuse to --exposure-sigma=.1 --saturation-weight=0 --exposure-mu=.35 and often enough that hits the spot :) Kay -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
