"Exposing to the right" is over exposed, what now?
in [Photoshop]
|
From: Clues on 8 Oct 2009 02:13 On Wed, 07 Oct 2009 17:19:12 -0500, John Sheehy <JPS(a)no.komm> wrote: >Exposure to the right is usually given as advice for RAW shooting. The >RAW data usually clips well above the level where a JPEG would clip it, Very true, for all those cameras that intentionally do a poor job at the RAW to JPG conversion. This is why RAW became so popular. DSLRs are inherently poor at this process. In fact it's the very reason that having access to the RAW data became so popular. Far too many DSLR owners wanted to repair in their resulting images what the DSLR's firmware programmers failed to do correctly in the first place. Companies then realized that they could turn this into an asset. Not only sell a camera that didn't work quite right, bundled with a kit-lens that wouldn't provide images any better than a Barbie-Cam, conning them into buying a $5,000-$10,000 lens to make their $500 DSLR functional, but now also sell them expensive editing software to fix what their cameras weren't doing properly to provide a useful image right out of the box. A popular business model for the last couple of decades. Not unlike how microsoft popularized this practice. They released Win95 in alpha phase (I was a pre-beta-tester and read the in-house memos back then). M$ was worried that they'd get sued worldwide for false-advertising if it wasn't released on the promised date. In fact they were already behind several deadlines and lawyers were starting to circle, ready to get their piece of the M$ castle. M$ decided to go ahead and ship the alpha-phase Win95 anyway. What buyer would ever know. They then also charged (on a per-call basis) their customers who bought this unfinished software for the tech-support needed to get it running somewhat. A very effective means to increase the profitability of an incomplete or improperly functioning product. (This is why you should never buy any windows platform until after the second or third service-pack is released. Let all the others happily and foolishly pay for the privilege to be microsoft's beta-tester grunts. When the reverse should be true. M$ should pay everyone to try out the first few issues of their latest versions. If they're lucky, and they pay all their beta-testers enough for all the time and work involved, we might even tell M$ what is wrong.) Sell them a DSLR that should work well and produce a useful image, but doesn't. Then provide a $5,000-$10,000 lens and $200 software that they'll need to get it working right. It's just the M$ marketing strategy applied to cameras, nothing more. Your valuable time that you spend in editing all your RAW images for this privilege is merely a courtesy feature and should go unmentioned, lest people become even more angry over how they've been cunningly turned into more CEOs' saps. There are many cameras now on the market where you will find little need for RAW. Take the RAW data and try to get it to look better than the JPG from the camera and you'll be hard pressed to better it. They manage to do the process right the first time with no need for time-consuming RAW editing intervention, other than setting white-balance and exposure properly while shooting, as any decent photographer will and should. Yes, they could easily provide the JPG images in the 16-bit HD-Photo format (nearly all good editors fully support it) with compression ratio saving options (as all editors do), for those of you that understand the need for the extra data when trying to recover from your snapshooters' mistakes and accidents. But then how could they sell the required RAW editing software to you too? Not going to happen if they can find a simple way to make you pay more of your money and waste even more of your valuable time. Enjoy your crippled cameras. You got exactly what you were willing to pay for.
From: DRS on 8 Oct 2009 02:23 "Paul Furman" <paul-@-edgehill.net> wrote in message news:hajt19$8gj$1(a)news.eternal-september.org > DRS wrote: [...] >> If the histogram at -4 is in fact more accurate than the histogram >> at 0 (the Canon default) then the photographer in fact has greater >> room for exposure compensation to the right than the histogram at >> contrast setting 0 indicates. For those wishing to compensate to >> the right this is useful information. What the photographer >> *subsequently* chooses to do to the exposure is another discussion >> entirely. > > What about saturation? Reducing saturation seems to give more detail > to the edges also. That is possible but not something for which I have evidence. My concern here was to short-circuit this increasingly hysterical subthread by demonstrating that Floyd has been tilting at windmills. Everything Doug McDonald has said about the effect on the histogram of changing the contrast setting on the 30D is borne out by the images on the page to which I linked. Since all the major DSLR manufacturers, to the best of my knowledge, derive histograms from the JPEGs then what he has described would be universally applicable.
From: Floyd L. Davidson on 8 Oct 2009 02:58 "DRS" <drs(a)removethis.ihug.com.au> wrote: >"Paul Furman" <paul-@-edgehill.net> wrote in message >news:hajt19$8gj$1(a)news.eternal-september.org >> DRS wrote: > >[...] > >>> If the histogram at -4 is in fact more accurate than the histogram >>> at 0 (the Canon default) then the photographer in fact has greater >>> room for exposure compensation to the right than the histogram at >>> contrast setting 0 indicates. For those wishing to compensate to >>> the right this is useful information. What the photographer >>> *subsequently* chooses to do to the exposure is another discussion >>> entirely. >> >> What about saturation? Reducing saturation seems to give more detail >> to the edges also. > >That is possible but not something for which I have evidence. My concern >here was to short-circuit this increasingly hysterical subthread by >demonstrating that Floyd has been tilting at windmills. Everything Doug >McDonald has said about the effect on the histogram of changing the contrast >setting on the 30D is borne out by the images on the page to which I linked. >Since all the major DSLR manufacturers, to the best of my knowledge, derive >histograms from the JPEGs then what he has described would be universally >applicable. The web page you cited has absolutely *nothing* on it that supports what you or McDonald have said. It talks only about the effect of changing contrast settings on the resulting JPEG image, not on the exposure or how it affects camera raw data. The histogram images show that *no* change in exposure is demonstrated, and that *logically* it is not possible to improve the accuracy of the histogram. (Indeed, the histograms shown are apparently purposely styled, with a white foreground against a *grey* background and with a grey scale at the bottom so that the lower right corner has a white border against the white foreground. They have made it impossible to see where the right edge of the data is!) What it does show is that the peaks in the graph get moved around, and if you *mistakenly* believe that where the peak is at is related to exposure, you will mistakenly think that contrast has somehow helped. The problem with saying that contrast adjustments make it more obvious where the right edge of the graph is at is that different setting help or hinder with different types of scenes. For one scene you might well be right that it is easier to see with low contrast, but just the same for the next scene you shoot it could be high contrast that would help. It is *not* inherently better for all purposes when set to low contrast! Showing that it helps to see the edge in one condition without being aware that the condition is a special case, leads to a logically invalid conclusion. -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com
From: DRS on 8 Oct 2009 03:20 "Floyd L. Davidson" <floyd(a)apaflo.com> wrote in message news:87zl82qsmh.fld(a)apaflo.com > "DRS" <drs(a)removethis.ihug.com.au> wrote: [...] >> That is possible but not something for which I have evidence. My >> concern here was to short-circuit this increasingly hysterical >> subthread by demonstrating that Floyd has been tilting at windmills. >> Everything Doug McDonald has said about the effect on the histogram >> of changing the contrast setting on the 30D is borne out by the >> images on the page to which I linked. Since all the major DSLR >> manufacturers, to the best of my knowledge, derive histograms from >> the JPEGs then what he has described would be universally applicable. > > The web page you cited has absolutely *nothing* on it > that supports what you or McDonald have said. It talks > only about the effect of changing contrast settings on > the resulting JPEG image, not on the exposure or how it > affects camera raw data. That is because they are two different things which nobody has conflated but you. The images on that site show exactly what Doug McDonald claimed and nothing more. Of course they show nothing on the effect on exposure or on Raw data but that was never the claim except in your mind. You have created a straw man. > The histogram images show that *no* change in exposure > is demonstrated, and that *logically* it is not possible > to improve the accuracy of the histogram. It is possible to improve the accuracy of the histogram relative to the Raw data and the fact that you cannot see it when it is right before your eyes is your problem, not ours. Those images show 5 different histograms where the only variable is the contrast setting. Since the exposure has not changed then only one of the histograms can be the most accurate relative to the Raw image. It happens to be the -4 setting. [...] > What it does show is that the peaks in the graph get > moved around, and if you *mistakenly* believe that where > the peak is at is related to exposure, you will > mistakenly think that contrast has somehow helped. Straw man. All five histograms have different right-edges, indicating different possibilities for exposure compensation before clipping occurs. One of the 5 most accurately indicates the true feasible extent of ETTR, and that one is the -4 histogram. That is all that has been claimed. > The problem with saying that contrast adjustments make > it more obvious where the right edge of the graph is at > is that different setting help or hinder with different > types of scenes. For one scene you might well be right > that it is easier to see with low contrast, but just the > same for the next scene you shoot it could be high > contrast that would help. It is *not* inherently better > for all purposes when set to low contrast! The question of when ETTR is suitable is a different question again., and is not something I addressed. > Showing that it helps to see the edge in one condition > without being aware that the condition is a special > case, leads to a logically invalid conclusion. You are really not in any position to lecture on logic.
From: Floyd L. Davidson on 8 Oct 2009 03:37
"DRS" <drs(a)removethis.ihug.com.au> wrote: >"Floyd L. Davidson" <floyd(a)apaflo.com> wrote in message >news:87zl82qsmh.fld(a)apaflo.com >> "DRS" <drs(a)removethis.ihug.com.au> wrote: > >[...] > >>> That is possible but not something for which I have evidence. My >>> concern here was to short-circuit this increasingly hysterical >>> subthread by demonstrating that Floyd has been tilting at windmills. >>> Everything Doug McDonald has said about the effect on the histogram >>> of changing the contrast setting on the 30D is borne out by the >>> images on the page to which I linked. Since all the major DSLR >>> manufacturers, to the best of my knowledge, derive histograms from >>> the JPEGs then what he has described would be universally applicable. >> >> The web page you cited has absolutely *nothing* on it >> that supports what you or McDonald have said. It talks >> only about the effect of changing contrast settings on >> the resulting JPEG image, not on the exposure or how it >> affects camera raw data. > >That is because they are two different things which nobody has conflated but >you. The images on that site show exactly what Doug McDonald claimed and >nothing more. Of course they show nothing on the effect on exposure or on >Raw data but that was never the claim except in your mind. You have created >a straw man. It may show what Doug is talking about, but it shows that it has nothing to do with correct exposure, which *is* the point of this discussion. You continue to say that effect on exposure in not significant, but the *fact* is that is the *only* thing that would be significant. You *cannot* get more accurate exposure without making the histogram more accurate. Contrast settings change the shape of the graph between the right most edge and the the left edge *but* *do* *not* *change* *the* *location* *of* *the* *right* *edge*. Hence if the right edge is set correctly with high contrast or low contrast, either way it is exactly the same exposure and neither is more accurate than the other. You continue to say that low contrast makes it easier to see where the edge is, but that is only true for special cases, and for an equal number of special cases high contrast would make it easier! >> The histogram images show that *no* change in exposure >> is demonstrated, and that *logically* it is not possible >> to improve the accuracy of the histogram. > >It is possible to improve the accuracy of the histogram relative to the Raw >data and the fact that you cannot see it when it is right before your eyes >is your problem, not ours. Those images show 5 different histograms where >the only variable is the contrast setting. Since the exposure has not >changed then only one of the histograms can be the most accurate relative to >the Raw image. It happens to be the -4 setting. All five show *exactly* the same exposure. I assume you are looking at the large peaks of the graph as they move more towards the right edge. That of course is *not* a measure of exposure. If the graphs were designed to show the effect they would also have to show histograms for some lower exposure and at some higher exposure, to demonstrate what one has to adjust for. > >[...] > >> What it does show is that the peaks in the graph get >> moved around, and if you *mistakenly* believe that where >> the peak is at is related to exposure, you will >> mistakenly think that contrast has somehow helped. > >Straw man. All five histograms have different right-edges, indicating >different possibilities for exposure compensation before clipping occurs. >One of the 5 most accurately indicates the true feasible extent of ETTR, and >that one is the -4 histogram. That is all that has been claimed. They all have exactly the same right edge, to the degree that it can be determined (which is exceedingly difficult because the images were designed purposely to de-emphasize that and to emphasize the changes that contrast actually does cause. That of course is perfectly reasonable because that is the topic they are discussing, and is the reason they do not mention exposure at all. >> The problem with saying that contrast adjustments make >> it more obvious where the right edge of the graph is at >> is that different setting help or hinder with different >> types of scenes. For one scene you might well be right >> that it is easier to see with low contrast, but just the >> same for the next scene you shoot it could be high >> contrast that would help. It is *not* inherently better >> for all purposes when set to low contrast! > >The question of when ETTR is suitable is a different question again., and is >not something I addressed. The above has nothing at all to do with suitability of ETTR, it is not something I've addressed either. Why are you? >> Showing that it helps to see the edge in one condition >> without being aware that the condition is a special >> case, leads to a logically invalid conclusion. > >You are really not in any position to lecture on logic. Other than I use it, and you don't? Seems like a good position... -- Floyd L. Davidson <http://www.apaflo.com/floyd_davidson> Ukpeagvik (Barrow, Alaska) floyd(a)apaflo.com |