[quote name='Brightcolours' timestamp='1286210378' post='3478']
We are not looking at different things. I seem to be looking at what we can see, you seem to not be doing that.
We see much more detail all over, and yes, we see some colour artifacts from a bad RAW converter. We were not discussing the colour, though, we were discussing the extra "detail", which is very apparent (it is in luminance, not in colour space... both get extracted from the RAW data in different ways).
[/quote]
Right, so these are the artifacts from a bad RAW converter and not the interpolation blocks, why I didn't think of that? Care to elaborate on which RAW converter produces such ugly artifacts so I could avoid it in the future? :-)
As I get it, you're basically refusing to see what I'm pointing at. Also I don't really understand why you always emphasize that detail is equal to luminance and chroma has nothing to do with that, when it's simply not true and shows [quote] a lack of insight in the matter [unquote].
I'm not talking about the colour accuracy here, but any chroma based details. Yes, these are called *details*, otherwise all B/W images would have exactly the same amount of details as the colour images. Or are you going to argue with this?
My point still stands, especially when your link very conveniently demonstrates an increase of sharpness on B/W sections of the photos only ;-)
[quote name='Lomskij' timestamp='1286270583' post='3490']
Right, so these are the artifacts from a bad RAW converter and not the interpolation blocks, why I didn't think of that? Care to elaborate on which RAW converter produces such ugly artifacts so I could avoid it in the future? :-)
As I get it, you're basically refusing to see what I'm pointing at. Also I don't really understand why you always emphasize that detail is equal to luminance and chroma has nothing to do with that, when it's simply not true and shows a lack of insight in the matter .
I'm not talking about the colour accuracy here, but any chroma based details. Yes, these are called *details*, otherwise all B/W images would have exactly the same amount of details as the colour images. Or are you going to argue with this?
My point still stands, especially when your link very conveniently demonstrates an increase of sharpness on B/W sections of the photos only ;-)
[/quote]
Nonsense, it shows that detail is in luminance (dark and light differences). That black and white are extremes in luminance information is a given.
Now, instead of digging your heels in the sand, actually LOOK at the samples.
The 20 dollar bill. Is it black and white? Of course not. It is different gradations of green (not pure green, check that in PS if you like). We see a big increase in detail. The cable next to it. Metal strands against a saturated blue background. Big increase in detail. The brass screw part of the cable, it is a yellow in different shades. Big increase in "sharpness".
The dollar bill "head", Has a light blue bird figure next to it. Big increase in the feather pattern "sharpness".
The face itself, only browns (so all colours in different gradations). Clearly has much more detail/sharpness.
The car's dials. The red hans clearly are much "sharper", we see very clear aliasing stair stepping in them. You refuse to see that.
The chrome ball in the lever. It shows more "definition" everywhere, including where we have blue, red. The steering wheel also has all colour channels in different values, it is not black and white at all. Yet we see the shades of brown. blue clearly and see detail increase in equal amounts.
The brass key. The brown fake wood.
It all shows the same increase in sharpness, in aliasing, in stair stepping. That you want to keep saying it is only in black and white areas is your choice, it does not make it true.
Yes, detail is in luminance information.
10-05-2010, 10:09 AM
(This post was last modified: 10-05-2010, 10:10 AM by Brightcolours.)
[quote name='genotypewriter' timestamp='1286245329' post='3482']
I did, in my previous post. Let's look at this another way... keeping the fact that some AA-filters are overly strong, are you saying that Bayer CFA manufacturers don't want people to get more detail out of their sensors?
The technical correctness of that test is questionable. Here are the reasons:
1. They're comparing two different cameras here. One is modified with a new clear glass filter which most probably has a different thickness to what the standard camera has. How accurately did they re-shim their focusing screen? And did they check the shimming of the stock camera's focusing screen to ensure that both produce the same level of accuracy (after all, manufacturing tolerances are normal).
2. These cameras don't have LV, so super-precise focusing necessary for focusing distances and magnifications like that is impossible... plain and simple. With or without eye-based focus bracketing, which is not accurate.
3. The AA-filterless camera here is what they refer to as a "hot rod" ("HR" postfix in the names, etc.) which is a full spectrum camera. Anyone who has experience with full-spectrum and IR imaging (/me raises both hands) knows that this introduces too many extra variables here.
3a. For one thing, most lenses focus UV, visible and IR light in to slightly different places (like multi-spectral CA) and only visible light can be seen at the time of focusing (obviously) but the final image contains UV and IR too.
3b. Then does the ICF they're using on the modified camera have the same spectral transmission curve as the ICF in the unmodified camera? Short answer: No. So what are we really seeing here (refer to 3a)? And on a side note, how did they manage to match the colours between the two so accurately when these image-capturing differences exist.
4. Not all AA filters are made equally. This is talking about a Nikon and a D200 of all things... technology has come a long way since then. Also according to what science is this observation going to be true for any other camera model? AA filters are different on every sensor. Since the D200 was available, for example, Canon has had the 5D (12MP FF), 30D (8MP), 40D (10MP), 450D (12MP), 50D (15MP), 60D (18MP), 1D2 (8MP APS-H), 1D3 (10MP APS-H), 1D4 (16MP APS-H), 5D2 (21MP FF), 1Ds3 (21MP FF/AA filter diff. from 5D2) all of which have their own unique AA filter. That's 11 different AA filters. Are you suggesting that all 11 of these were made like the D200's AA filter?
For me the biggest weakness in this test is what I mentioned in #2 above. Don't forget that these maxmax guys are also trying to sell their service...
GTW
[/quote]
1. Maxmax does use a correct replacement glass with the same refractive index.
2. That is just plain silly. The silliest point you can have made... sure, blame it on a wrongly focussed non-HR-ed image. Too silly for words. But you can look at the D300 and D700 samples. Or any other samples of people who have had their camera converted to AA-filter-less.
3. It is NONSENSE what you say. Maxmax uses HR for AA-filterless. High Resolution. Hot Rod visible.
They also do OTHER conversions, which are called IR-only (as they state: Dedicated IR cameras. 715nm or 830nm options. Same price.). And High Red. And UV+Visible+IR. And UV-ONLY. And Black and White (CFA removal).
4. I know AA-filters all are different. What is your point here? Sigh. Yes, I happened to show an example of the difference with a D200. So what? You go find comparisons of any Canon model with and without AA-filter if that makes you feel better.
It does not matter how strong the AA-filter is, what matters is that we can see foveon like detail (AA-filter-less detail) and that we can clearly see that luminance detail is clearly extracted from the different colour sensels.
Does that mean the output will be the same as what a Foveon sensor outputs? No of course not. I never said anything like that either. I merely have explained that a foveon/sigma sensor appears to resolve more is because they decide to not use an AA-filter. And we can see that clearly, we see (with the D200 without AA-filter) pixel wide lines, sharp aliased lines.
For some reason lomskij things black and white makes some exception.. I have no idea why. Since black and white are the clearest example how luminance information gets extracted very detailed from the different colour sensels.
[quote name='Brightcolours' timestamp='1286272202' post='3492']
Nonsense, it shows that detail is in luminance (dark and light differences). That black and white are extremes in luminance information is a given.
Now, instead of digging your heels in the sand, actually LOOK at the samples.
The 20 dollar bill. Is it black and white? Of course not. It is different gradations of green (not pure green, check that in PS if you like). We see a big increase in detail. The cable next to it. Metal strands against a saturated blue background. Big increase in detail. The brass screw part of the cable, it is a yellow in different shades. Big increase in "sharpness".
The dollar bill "head", Has a light blue bird figure next to it. Big increase in the feather pattern "sharpness".
The face itself, only browns (so all colours in different gradations). Clearly has much more detail/sharpness.
The car's dials. The red hans clearly are much "sharper", we see very clear aliasing stair stepping in them. You refuse to see that.
The chrome ball in the lever. It shows more "definition" everywhere, including where we have blue, red. The steering wheel also has all colour channels in different values, it is not black and white at all. Yet we see the shades of brown. blue clearly and see detail increase in equal amounts.
The brass key. The brown fake wood.
It all shows the same increase in sharpness, in aliasing, in stair stepping. That you want to keep saying it is only in black and white areas is your choice, it does not make it true.
Yes, detail is in luminance information.
[/quote]
Ok. I guess it will be easier if I go step by step:
1. "B/W sections" - my bad, what I was actually trying to say, is that those sections demonstrate luminance only based details, with chroma having little or no effect on the image contrast. Now it sounds more or less right.
2. So just to reiterate - according to you, only luminance (dark and light difference) creates detail, while chroma (wavelength difference) does not, correct?
3. I'm looking into these images again, and what I can see quite clearly that if there's a substantial change in brightness but not in colour (i.e. shades of green, brown, whatever), it gets sharpened quite neatly with just a minor increase in aliasing "stepping" - dollar bill, digits on a speedometer, etc. Except maybe the leather steering wheel, which looks like a perfect "snap to grid" example to me.
However all details created by a strong change in colour, e.g. edges of a speedometer needle, red belt, brass screw and blue keyfob, all red reflections on chrome surfaces, metal strands against blue background, etc. show interpolation artifacts and much larger "stepping". Please don't tell me you don't see it :-)
BTW, [quote] The car's dials. The red hans clearly are much "sharper", we see very clear aliasing stair stepping in them. You refuse to see that. [unquote] - that's my point exactly!
All in all, it seems that this dispute successfully got itself into a deadlock. It would be really helpful to get some raw files both from Sigma and filter-less Bayer, unfortunately I can't find it online. Genotypewriter - you mentioned that you have a 50D with AA filter removed, would you mind sharing with some raw files please?
[quote name='Brightcolours' timestamp='1286273352' post='3493']
For some reason lomskij things black and white makes some exception.. I have no idea why. Since black and white are the clearest example how luminance information gets extracted very detailed from the different colour sensels.
[/quote]
That's pretty simple - to my understanding, image detail is a result of combined luma and chroma data. And the only situation when errors in the chroma processing (interpolation for instance) do not affect the detail is in monochrome scene. I thought I've explained my point of view quite clearly in my previous posts.
All in all, it seems that this dispute successfully got itself into a deadlock. It would be really helpful to get some raw files both from Sigma and filter-less Bayer, unfortunately I can't find it online. Genotypewriter - you mentioned that you have a 50D with AA filter removed, would you mind sharing with some raw files please?
[/quote]
Here you can find a 5D without AA-filter and with AA-filter RAW file.
http://www.maxmax.com/hot_rod_visible.htm
Again, regarding our discussion:
I do not claim ANYWHERE that a bayer CFA AA-filterless sensor will get the SAME results as a Sigma foveon withouth AA-filter.
I claim that extra "detail" from the foveon is mainly to do with the AA-filter lack, and that it mainly concerns false detail. And that detail is luminance based, and luminance data very successfully gets extracted from the bayer-CFA RAW data (as can very clearly be seen from the D200 examples).
My God, you pixle peepers make my head spin. There are a lot of assumptions in every entry here. Why don't you take a look at the interview with he heads of Sigma and Foveon showing lots of technical facts just published on DPReview. Maybe you can get some facts to discuss. Kindly Vieux Loup
[quote name='Vieux loup' timestamp='1286295028' post='3504']
My God, you pixle peepers make my head spin. There are a lot of assumptions in every entry here. Why don't you take a look at the interview with he heads of Sigma and Foveon showing lots of technical facts just published on DPReview. Maybe you can get some facts to discuss. Kindly Vieux Loup
[/quote]
Don't take the blurb from Sigma purely as facts.
They do not even talk about how they do NOT actually have a layer that captures green, red and blue.
Just the same marketing blurb we have heard before, trying to downplay the competition and discard the downsides from their own approach.
If that is your idea of technical facts...
I like these fights/discussions, and especially when Geno trows a bumernags time to time al thway from Melborne.:-)
genotypewriter
Unregistered
[quote name='Brightcolours' timestamp='1286273352' post='3493']
2. That is just plain silly. The silliest point you can have made... sure, blame it on a wrongly focussed non-HR-ed image. Too silly for words. But you can look at the D300 and D700 samples.
[/quote]
In science we don't use words like "silly" to get our points across. It just goes to show that you have little knowledge and exposure you have in the scientific method.
[quote name='Brightcolours' timestamp='1286273352' post='3493']
Or any other samples of people who have had their camera converted to AA-filter-less.
[/quote]
To put it in your sort of words... can't you read? I already have an AA-filterless camera and I have used Foveon cameras. You're the one who is speculating everything here without experience anything. Have you used any of these things you're talking so confidently about? No. lol
[quote name='Brightcolours' timestamp='1286273352' post='3493']
3. It is NONSENSE what you say. Maxmax uses HR for AA-filterless. High Resolution. Hot Rod visible.
[/quote]
Manners aside, I guess you can't even read... if it's a visible only camera why do they use an ICF filter then as they describe in the first paragraph? lol
[quote name='Brightcolours' timestamp='1286273352' post='3493']
I merely have explained that a foveon/sigma sensor appears to resolve more is because they decide to not use an AA-filter. And we can see that clearly, we see (with the D200 without AA-filter) pixel wide lines, sharp aliased lines.
[/quote]
The only thing you explained is your uninformed guess, which is also scientifically unfounded. Demosaicing reduces spatial resolution. Don't you know how that works? Bayer puts 3 points in to 1 coloured pixel, Foveon keeps 3 separate coloured pixels. This is where I stop explaining because the other party is not willing to learn... at least read that dpreview article that was posted recently on the SD1. But I doubt you'd understand the theory.
GTW
|