E5 sharper than E3? Not by much.

[rovingtim]

You are still ignoring the resolution measured in his links, do you agree or disagree with their resolution measurements. If you disagree why?

I'll say it again:

When I was working with the crops to post with Mr Cat's crops, I couldn't get the E5's 200% crop of the print to be as clear as the Raw Therapee's rendition no matter what I did. On the other hand, I could bring out other detail that RAW Therapee was missing, especially in the E3 shot. In other words, from my experience it seems RAW Therapee is better suited to the E5 than the E3. So, person who sees 'deeper agenda's', what does a higher MTF score mean? What is it measuring?

We cant see that, so we have to take your word for it.

No you don't. Download the ORF's and try it yourself.

 

[rovingtim]

I should have anticipated this misunderstanding.

The reason I didn't use Lightroom on the E5 was because when I did in a previous thread, a number of people showed up and told me that my results were invalid because I was using Lightroom on the E5 and Lightroom doesn't have good algorithms for the E5.

So, in order not to have that same charge leveled at me again, I didn't. Now I get accused of not using Lightroom on both cameras and that invalidates my findings. Ironic, eh?

However, the interesting upshot of all this is that in playing with the different RAW developers (and comparing crops to Mr Cat's) I have found different RAW developers work differently on the different cameras.

Raw Therapee does an excellent job pulling detail out of an E5 file: in the printed areas, I can't equal it using Lightroom. However, I can get much more detail out of the E3 file using Lightroom than are showing in Mr Cat's E3 crops.

So, it gets interesting. What is equal?

It also got me wondering about the validity of the mtf tables (which I have previously taken for gospel).

 

[rovingtim]

See my reply to D3xmeister.

 

[rovingtim]

When I did my first comparison, a group of fine individuals said my finding were invalid because I was using Lightroom on the E5 and Lightroom had poor algorithms for the E5.

As such, my findings were invalidated.

As a direct response, I didn't use Lightroom on the E5. Now I'm being told that my findings are invalid because I didn't use Lightroom on the E5.

Do please tell me how I can possibly do right by this forum?

 

[matt_j]

It is quite stimulating.

I feel that the E-5 like many new cameras is somewhat overhyped (following great disappointment with its specs at time of release). Many people, or should I say guys have a strong need to own and use stuff imbued with fantastical qualities.

This thread demonstates that there is no consensus about processing Olympus' files (Lightroom / Viewer 2 preference fluctuates from thread to thread).

I believe that the E-5 has a potential edge sharpness-wise. Just as camera X having a dynamic range over camera Y - one can use it to process files more freely. This advantage manifests itself only in demanding conditions, and let's face it - most prosumers like many of us are hadly ever faced with such conditions (print landscape or portraiture shots that big).

And a big thumbs up from me for Tim for defending his initial post all the way down here. Do challenge him. I got me some crisps

--
http://www.digital-photography.pl

 

[odl]

holding baby typing with 1 hand awkward.

there is more detail in E5, see bottle labelcrop, some small artifacts, but crisp detail. what we expext.

with increase allowance for detail, fabric looks better. reds more in lline i did no colour pp for match, only for detail need to boost blacks a bit, noise also looks very good.

quik pp at 3am













--
“Imagination is more important than knowledge. For
knowledge is limited to all we now know and
understand, while imagination embraces the entire
world, and all there ever will be to know and
understand.” - Albert Einstein

 

[rovingtim]

holding baby typing with 1 hand awkward.

there is more detail in E5, see bottle labelcrop, some small artifacts, but crisp detail. what we expext.

Careful of false detail on the E5 shot, the most obvious being the horizontal lines in the pink cloth.

However, I do want to stress that I am not saying the E5 doesn't have more detail, only that the detail increase has been grossly exaggerated. 'Lifting a veil' from the lenses, etc.

quik pp at 3am

Mr Cat has demonstrated that RAW Therapee is better at pulling high contrast detail out of E5 files (like black print) than ACR. I can't get ACR anywhere close to the clarity he's getting in those areas with us both using the same E5 orf.

However, I feel E3 + Lightroom can better the RAW Therapee E5 file -- in certain places -- but the E5 does have more overall detail, however, the difference isn't much.

 

[Rriley]

You do go on.

He has made a very valid point using external and accepted information.

He's been banging on about me using highly magnified crops to point out moire. That was what that was about. The clue is my next statement in the post to him:

Why moire at huge magnifications is important:

http://forums.dpreview.com/forums/read.asp?forum=1022&message=38581467

So, odl, what is your 'deeper agenda' for not noticing this?

Wasnt this thread about sharpness? Your point was the E-5 wasnt that much sharper than the E-3, unless you dont think Rileys linked tests are reliable (please explain why if not) then he has shown conclusively something which confirms what E-5 owners have been saying.

It's interesting that you bring this up because I was thinking about the mtf's just before I logged on. Are they taken before sharpening? Which demosiacing engine do they use?

DxO's RAW

in any event sharpening in jpeg (not that this appears to be the case) shouldnt assist an MTF50, its more likely to reduce it

MTF profiles are provided in R, G, and B channels for RGB images and R, Gr, Gb and B for RAW images, measured at four different locations. Measurements are more precise for low frequencies than high frequencies.

When I was working with the crops to post with Mr Cat's crops, I couldn't get the E5's 200% crop of the print to be as clear as the Raw Therapee's rendition no matter what I did. On the other hand, I could bring out other detail that RAW Therapee was missing, especially in the E3 shot. In other words, from my experience it seems RAW Therapee is better suited to the E5 than the E3. So, person who sees 'deeper agenda's', what does a higher MTF score mean? What is it measuring?

well I did supply you with a link that answers more fully what you question now
in any event its a pretty good primer
http://www.diwa-awards.net/wip4/modulation-transfer-function-mtf/d.epl?id=156093

Modulation Transfer Function (MTF)

Image resolution is defined by the number of picture elements (pixels) on the image sensor. More pixels will, in principle, make smaller details distinguishable in the image. A high resolution sensor can "see" finer details as the pixels are packed closer together on the sensor. However, high sensor resolution requires a lens with adequate "resolving power", so that those fine details can be recorded by the sensor. Sharpness (see "Blur" test") and optical resolving power (MTF) are vital for high image quality.

What is Modulation Transfer Function?

Imaging is simply a matter of contrasts. Any pattern consists of contrasts, usually with sharp edges. When contrast lines in the picture are clear, the image is perceived as sharp, while fuzzy contrast lines make the image look blurry. However, our eyes also tend to recognize sharpness by the level of contrast. Fuzzy lines with high contrast may appear sharper than fine lines with low contrast. Therefore, lens manufacturers try to optimize both criteria, high resolving power and high contrast

The limit for a sensor's ability to distinguish contrast lines is fairly easy to understand. Any given pixel on the sensor can only define one gray level at a time. Therefore, a minimum of two pixels is necessary to define a contrast edge (one dark and one bright pixel next to each other), hence defined as "line pairs". The restriction is the distance between two adjacent pixels. This fact is described as the "Nyquist theory". For instance, the Nyquist frequency of an APS size sensor (22.5x15 mm) with 8 MP resolution (3504x2336 pixels) is 78 line pairs per millimeter (lp/mm).

I think if you had serious concerns about what you say, you should be all over this

same lens on both E3 and #5





--
Riley

any similarity to persons living or dead is coincidental and unintended

 

[rovingtim]

I didn't fully understand mtf. That explanation goes back to the first thread: if the E5 is generating an abundance of false detail, the mtf will be higher won't it?

Or am I missing something?

Edit: (if I'm not missing something, maybe this is another reason to go for light AA's ... the mtf shoots up.)

 

[Rriley]

I didn't fully understand mtf. That explanation goes back to the first thread: if the E5 is generating an abundance of false detail, the mtf will be higher won't it?

no

but you rightly use the operative 'if', so the better question might be 'if'. Ive certainly seen that in vegetation, but not commonly or anything like that

Or am I missing something?

no, the detail needs to be consistent, what you describe means it would go to indistinct mush earlier as you look for finer detail

Edit: (if I'm not missing something, maybe this is another reason to go for light AA's ... the mtf shoots up.)

well exactly, and i expect others will (as usual) follow,
the key differentiator then is, the lenses

--
Riley

any similarity to persons living or dead is coincidental and unintended

 

[Melvinphoto]

This test highly depends on your post processing skills and methods used. For a sharpness test, compare the RAW files unprocessed at 100%. The difference between E5 and E3 is huge in favour of the E5.

You can say whatever you like about the E5, I think 99,9% of E5 owners find E5 to be much sharper than any other 4/3rds camera incl. E3

--
http://www.redeker-photography.com

 

[rovingtim]

This test highly depends on your post processing skills and methods used.

Indeed, as do good photographs.

For a sharpness test, compare the RAW files unprocessed at 100%. The difference between E5 and E3 is huge in favour of the E5.

In unprocessed RAWs at 100% I wouldn't disagree. However, that is not the end of the story.

Comparing the torque of a tractor engine with a sports car and concluding the tractor must therefore accelerate better is not the a good way of determining the whole story. If you don't put all the pieces together, you don't get the full picture.

You can say whatever you like about the E5, I think 99,9% of E5 owners find E5 to be much sharper than any other 4/3rds camera incl. E3

This is not true. The L10 might have even had a lighter AA (I would love to see a direct comparison) and the E510 and E410's (and E400) tended towards lighter AA's as does the EPL2.

Is it much sharper? If it is it should be easy to show me up in a direct comparison shouldn't it? All we need is a file from each camera of the same thing and Imaging Resources has supplied exactly that. It didn't look 'much sharper' to me.

Feel free to download the file and prove me wrong. Otherwise its just your say so.

I'm still waiting. The challenge is easy to implement and only Mr Cat has taken me up on it.

 

[djbrom]

... you can take up my challenge I set in the OP.

For those who will now insist that I did this test wrong somehow and stacked the deck against the E5, feel free to process the E5 raw anyway you like and I will happily set it against the E3 file.

I have made comparisons myself in a real life test when i owned both cameras (with the same 12-60 lens with same settings) and noticed the difference regardless of how the files were processed.

Using the same RAW engine and calling it even may not be the whole story because from what I have seen of Mr Cat's renditions, I can do a better job with E3 files with Lightroom.

I think RAW Therapee really does work well for the E5 though.

I gotta say that you certainly striked curiousity in my little brain about the different raw converters. I have often heard about RAW Therapee and tried it a few times but i guess just went back to ACR (and the times I used Bibble) for convenience and the fact that i piad for it so i had better use it

But what you certainly proved previously is RT's ability to correctly render detail with the e5 orf's. Thankyou for that

Dont get me wrong, the e3 can pump out some detail if processed well and I have seen this in my own files. I dont need to take up your challange...i am happy with what I have seen with my eyes and testing.

No problem.

Having said that I have seen the comparison that someone else has done of the files on equal terms and again there is nothing more to say.

Equal terms ... are you sure?

(see my post to odl, the part about mtf scores: http://forums.dpreview.com/forums/read.asp?forum=1022&message=38590874 )

Tim, im not putting down the e3

I don't love the E3. I love the E1. Put down the E1 and I will become irrational and illogical.

lol. I miss my old e1. Thats one camera i should not of sold.

But thats not the discussion here, the ability to resolve detail is and it is clear that the e3 is not capable of producing the detail the e5 can and thats a known fact (due to aa filter).

I didn't say otherwise. What I'm objecting to is the sense the E5 kills the E3 in terms of detail. From the images I have seen posted here, that is not the case. I think the E5 is a bit better. At no time do I suggest the E3 matches the E5 in detail.

Sorry that I misunderstood your post. I should think more before I post.

But the e3 also does not exibit moire to the same degree that the e5 does.

Some people define moire is only rainbow patterns of colour, so I would like to add 'colour casts' and 'false detail' to that as well.

I have seen differences between all lenses I have (im obviously excluding lense i never used on the e3) which are the 12-60, 11-22, 50 & 50-200. There is a difference with these lenses and i could only imagine what the difference would be with the SHG's.

See msusc's crops. They are in the answer immediately after my OP.

Thanks for your comments. Much appreciated.

 

[Rriley]

This test highly depends on your post processing skills and methods used.

Indeed, as do good photographs.

For a sharpness test, compare the RAW files unprocessed at 100%. The difference between E5 and E3 is huge in favour of the E5.

In unprocessed RAWs at 100% I wouldn't disagree. However, that is not the end of the story.

Comparing the torque of a tractor engine with a sports car and concluding the tractor must therefore accelerate better is not the a good way of determining the whole story. If you don't put all the pieces together, you don't get the full picture.

You can say whatever you like about the E5, I think 99,9% of E5 owners find E5 to be much sharper than any other 4/3rds camera incl. E3

This is not true. The L10 might have even had a lighter AA (I would love to see a direct comparison)

you can just work it out, differences in resolution at 'The ROOT of (10/12) which Ive already done, and it isnt lighter BTW

and the E510 and E410's (and E400) tended towards lighter AA's as does the EPL2.

havent compared EPL2, all the others have less res

Is it much sharper?

well yes

If it is it should be easy to show me up in a direct comparison shouldn't it? All we need is a file from each camera of the same thing and Imaging Resources has supplied exactly that. It didn't look 'much sharper' to me.

Feel free to download the file and prove me wrong. Otherwise its just your say so.

I'm still waiting. The challenge is easy to implement and only Mr Cat has taken me up on it.

fcol how long does this bs have to go on before you concede

--
Riley

any similarity to persons living or dead is coincidental and unintended

 

[DonParrot]

The E-5 shot should be slightly larger:





(Edit: corrected this size comparison.)

Thanks for the explanation. But I still don't understand the reason for this difference in size. We have got two sensors of the same format, so how can the number of pixels used on this area make an impact on the size of the picture? Rather weird from a logical point of view, I'd say. So, where is the mistake im my thought chain?

My example above shows each sample resized to 16%: this means 100 pixels in either direction is represented by 16 pixels. The E-5 has more pixels, so it looks larger than the E-3 sample.

How this affects the 100% size crops (1 pixel in the sample represented by 1 pixel on the display) is an apparent magnification factor. The E-5 has 4032 horizontal pixels, and the E-3 3648. This yields a result from the E-5 that is 4032/3648 times as large. (The same holds true for the vertical direction as well, obviously.)

What this means for a print of the same size is that the E-5 has slightly more information available. If you print large enough the E-3 sample will start to look blurry (or fuzzy, or otherwise not right) at a smaller enlargement size than the E-5: in theory, the E-5 sample can be enlarged 4032/3648 times as much as the E-3 sample and retain the same sharpness. In practice, the E-5 yields more detail as well as more pixels, so it should be able to be enlarged further still.

Thanks again.

 

[DonParrot]

However, the interesting upshot of all this is that in playing with the different RAW developers (and comparing crops to Mr Cat's) I have found different RAW developers work differently on the different cameras.

Raw Therapee does an excellent job pulling detail out of an E5 file: in the printed areas, I can't equal it using Lightroom. However, I can get much more detail out of the E3 file using Lightroom than are showing in Mr Cat's E3 crops.

So, it gets interesting. What is equal?

That's why I have been claiming for years - not in this forum, haven't been here, that long - that comparing RAWs is BS as you don't compare what a camera can do but what a RAW converter can do and the PP skills of the person using it, combined with his/her knowlegde about the respective converter and - of course - his/her likings. BS!!! Even more so as - just as you said - one converter suits the one camera better, the other the other.

So if there were a camera delivering outstanding JPEGs but RAWS no converter would be able to deal with perfectly?
Good camera?
Bad camera?

And who is to blame? The camera manufacturer who opted for a path the comverters can't follow, or the converter manufacturers who aren't able to provide the tool it takes to properly PP the camera's RAWs.

I'm so tired of these useless debates. Why please can't I force myself to just shut it?!

And then we've got another point - which shouldn't be taken personal by anybody.

If somebody is keen on proving that the E-3 is nearly on par with the E-5 he will - that's just human - make a bigger effort to squeeze the maximum out of the E-3 RAWs than he will do regarding the E-5 RAWs. And those who want to prove the opposite will do vice versa. You just can't help it - it's part of our character. You may try as hard as you can, you just can't be objective if you want to prove a point. Indeed, I think you never can.

 

[FrankyM]

You said it yourself, Brent - it's too "less biased"...

 

[rovingtim]

If somebody is keen on proving that the E-3 is nearly on par with the E-5 he will - that's just human - make a bigger effort to squeeze the maximum out of the E-3 RAWs than he will do regarding the E-5 RAWs. And those who want to prove the opposite will do vice versa.

That's why my challenge is for other people to process the E5 while I process the E3. Then we compare.

 

[boggis the cat]

Here are your crops next to my attempts.

What do you think?

I think that you're missing the apples and oranges point.

The E5 is different from the E3 but you claim the E5 has better detail. You are comparing apples and oranges.

Apples and oranges applies to your weird insistence that using different post-processing is a valid methodology to compare detail. The second sentence of the partial quote is You can get better results from either camera with different post-processing.

It is your claim that the E-5 has no significant detail advantage over the E-3, beyond the extra pixel resolution. Your opening lines in the first post:

Right, keep in mind there is some detail in the E5 file that simply can't be captured by the E3 because it's missing 2 million pixels.

However, does the lighter AA make a huge difference? I would suggest not.

You use the same RAW developer on an apple and an orange and then claim it is equal. How so?

The processing is equal. If you wish to isolate camera differences then you cannot introduce other variables. You cannot determine an aspect such as relative detail by taking the samples and applying different processing .

My challenge is simple. Pick any demosiacing process and any pp process with you E5 and I pick whatever I want with the E3 and then we compare.

We already know that there are significant differences in what output different raw developers produce. If you are trying to compare the cameras then the raw development has to be kept the same -- you can't change variables that are unrelated to what you are trying to compare: you want to compare E-5 and E-3 detail.

In any case, even your own differently-processed results show that the E-5 delivers more detail.

By the way, I never said the E3 has as much or more detail as the E5; I said 'not by much'. It's in the title of this thread.

Your opinion is at odds with the majority, and in particular with those who own and use both models (I only have the E-5). You are effectively asking for a post-processing comparison and then claiming that this is directly related to the inherent qualities of the cameras. This is simply incorrect.

You can get better results from either camera with different post-processing. Why not apply a Gaussian blur to the E-5 result and claim that the E-3 has more detail? It would be just as valid.

Really? You are a strange Cat.

My point is that deliberately blurring one output would be just as valid as applying different post-processing: they're both an arbitrary decision and unrelated to what you are ostensibly trying to compare.

It is still pretty obvious in your unequal processing samples that the E-5 has more actual detail. Compare the samples to the Pentax 645D sample (that has the highest resolution available of that scene) and you can see the fabric detail quite easily.

I used the D3x, but you are right that the Pentax has more still. Since I had already posted the D3x crops to show the real cloth patterns as evidence of false detail in the E5 file, I didn't quite understand your condescending tone in your post in the other thread ...

http://forums.dpreview.com/forums/read.asp?forum=1022&message=38584987

Really. A good point? Well, good thing I had already made it then ... in the original comparison.

The D3x sample doesn't show the fabric detail as well as the sample from the medium-format Pentax 645D. From that sample we can see the structure of the weave in the fabric quite clearly, and thus have a much better idea of whether the E-5 is representing real detail correctly or simply supplying false detail.

My opinion is that the E-5 doesn't get it completely correct, but the E-3 doesn't show the fine detail at all. There is an element of "visual moire" involved, too, as the 100% and 200% crops I posted should illustrate. This isn't a surprise to me as I understand sampling and aliasing. In some situations the lack of detail may be preferable, but in general the greater detail of the E-5 is going to be the better option.

Recall that the general consensus was that the E-3 AA filter was too strong . Olympus has provided a weaker AA filter in the E-5 and put in some clever raw processing tricks to try to overcome moire. It appears that those tricks may work but your raw developer can still cause problems. Maybe a case can be made that the E-5 AA filter is too weak (or the moire combating algorithms are insufficient), but that case has certainly not been proven as yet.

In any event, it still remains that the E-5 has considerably more detail than the increased pixel count represents -- it out-resolves all other 12 MPixel Olympus bodies.

 

[boggis the cat]

However, the interesting upshot of all this is that in playing with the different RAW developers (and comparing crops to Mr Cat's) I have found different RAW developers work differently on the different cameras.

You must have missed my thread:
http://forums.dpreview.com/forums/read.asp?forum=1022&message=38561969

Raw Therapee does an excellent job pulling detail out of an E5 file: in the printed areas, I can't equal it using Lightroom. However, I can get much more detail out of the E3 file using Lightroom than are showing in Mr Cat's E3 crops.

Try applying the same processing to both samples. You'll find that you cannot get as much detail from the E-3 file . In fact, you can't get nearly as much.

(rriley clued you in to the MTF50 results using the same lens on the E-3 and E-5. Those tests are incontrovertible proof for the much greater detail that can be obtained from the E-5.)

So, it gets interesting. What is equal?

How about applying equal processing?

That's why I have been claiming for years - not in this forum, haven't been here, that long - that comparing RAWs is BS as you don't compare what a camera can do but what a RAW converter can do and the PP skills of the person using it, combined with his/her knowlegde about the respective converter and - of course - his/her likings. BS!!! Even more so as - just as you said - one converter suits the one camera better, the other the other.

Well this is certainly relevant to what rovingtim appears to believe is a valid methodology -- processing the files differently. Here are crops of the E-5 and E-3 samples using the same LR development:





There is a lot more detail in the E-5 crop, don't you think?

Now, here is the problem found with Lightroom (and most other developers): look at the fabric with the leaf pattern closely and you will see artefacts. RawTherapee "amaze" algorithm and Olympus Viewer do not produce such artefacts from the E-5 raw.

So if there were a camera delivering outstanding JPEGs but RAWS no converter would be able to deal with perfectly?
Good camera?
Bad camera?

And who is to blame? The camera manufacturer who opted for a path the comverters can't follow, or the converter manufacturers who aren't able to provide the tool it takes to properly PP the camera's RAWs.

I'm so tired of these useless debates. Why please can't I force myself to just shut it?!

The E-5 raw files seem to be able to be dealt with perfectly well by Olympus and also by a free product. I would argue that it is the non-performing software that is to blame.

And then we've got another point - which shouldn't be taken personal by anybody.

If somebody is keen on proving that the E-3 is nearly on par with the E-5 he will - that's just human - make a bigger effort to squeeze the maximum out of the E-3 RAWs than he will do regarding the E-5 RAWs. And those who want to prove the opposite will do vice versa. You just can't help it - it's part of our character. You may try as hard as you can, you just can't be objective if you want to prove a point. Indeed, I think you never can.

This is certainly true if you believe rovingtim's "methodology" is valid.

But it isn't.

The only valid methodology is to process the files the same way, then compare the results. Remember that this thread is an argument that the E-5 doesn't deliver as much detail over the E-3 as it in fact does. The method to "prove" this contention is to use different developers for each file then post-process the E-3 result further. Utterly nonsensical.