Shifting from 808 symbian to lumia 1020

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
I think there's more to that. Your interesting choice of words "he was caught" and attempted to "weasel his way out" seems rather deliberate to paint Damian as being dishonest or had something to hide. I invite you to explain how that is the case.

Maybe that was a bit harsh on my part. But I do think he gave an evasive answer. The question was very straightforward and to the point: "If we only needed a 2MP image, for instance, we would traditionally take a full resolution shot, and then downsample it with Photoshop. Does PureView provide downsampling algorithms superior to those of Photoshop or other image editing software?" His answer was very roundabout. Did he answer "yes" or "no"? My interpretation is that he tried to make the best of a difficult situation by effectively answering "no" in a way that still paints the PureView resampling in as positive a light as possible.

Good interviews often involve difficult and uncomfortable questions. It's the interviewer's job to cut through the hype and try to get a straight answer on things like this. A good interview brings new information to light instead of just rehashing what interested parties already know. Anyway, aside from being wrong about JPEG, I think Damian answered skillfully here, considering his position.

What I mean about JPEG... He said "But you're then handling the JPG file that was saved, so you're probably better off doing it at point of capture." But the question was framed in the context of the process the end user would go through to get the best possible image at a particular size. If you resample down from a full-res JPEG in Photoshop, you then have access to a clean, effectively unartifacted resampled image. But if you let the phone do the resampling, your resampled image now has the JPEG artifacts. Since the phone won't let you save a resampled image in uncompressed form, you have to choose the lesser of two evils. Which one does more damage to the final image? Using a JPEG-compressed image as input to the resampling, or saving the output as JPEG? I think it's pretty clear which way yields better results.
 

Bahamen

New member
Nov 20, 2012
272
0
0
Visit site
Maybe that was a bit harsh on my part. But I do think he gave an evasive answer. The question was very straightforward and to the point: "If we only needed a 2MP image, for instance, we would traditionally take a full resolution shot, and then downsample it with Photoshop. Does PureView provide downsampling algorithms superior to those of Photoshop or other image editing software?" His answer was very roundabout. Did he answer "yes" or "no"? My interpretation is that he tried to make the best of a difficult situation by effectively answering "no" in a way that still paints the PureView resampling in as positive a light as possible.

So you are certain there's only "Yes" or "No", and there's no possibility of "Maybe"? What if the answer is inherently subjective and can't be answered definitively?

What I mean about JPEG... He said "But you're then handling the JPG file that was saved, so you're probably better off doing it at point of capture." But the question was framed in the context of the process the end user would go through to get the best possible image at a particular size. If you resample down from a full-res JPEG in Photoshop, you then have access to a clean, effectively unartifacted resampled image. But if you let the phone do the resampling, your resampled image now has the JPEG artifacts. Since the phone won't let you save a resampled image in uncompressed form, you have to choose the lesser of two evils. Which one does more damage to the final image? Using a JPEG-compressed image as input to the resampling, or saving the output as JPEG? I think it's pretty clear which way yields better results.

I believe Damian meant that the 808's downsampling approach takes raw image data, before any JPEG compression, and uses that raw data for downsampling purposes. The alternative, as suggested by the interviewer, involves taking the 41/38/34MP JPEG (high res, but already compressed) file, and run it through Photoshop. Now, common sense dictates that downsampling directly from RAW data file is superior compared to downsampling from a high-res JPEG file. Whether or not this works well in practice is a separate matter, but Damian is certainly right that there are benefits to process an image directly from raw file as opposed to JPEG file.
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
Guys, I know the full-res images are compressed. And I know there's a benefit to using the uncompressed data as input to the resampling. What I said is that since the phone only lets you save as JPEG, you - the end user - have to choose one of the following paths if you want the highest quality image at a particular size:

Path 1: raw data -> compressed by phone as JPEG -> resampled by Photoshop/etc -> not necessarily compressed at this point, it's up to you
Path 2: raw data -> resampled by phone -> compressed by phone as JPEG

I think the first option will yield superior quality in normal usage scenarios. That is, I assume that with path 1 you will resize down by enough to clean up the JPEG artifacts present in the full-res image, which I think is a good assumption because you need to resize down substantially for other reasons to get good quality. And I assume that with path 2 you will not further resize the image down by enough to clean up its JPEG artifacts - if you did that, whatever benefit this path had when it comes to sampling uncompressed data as its input would be moot anyway. And finally I assume that the benefit that path 2 has of sampling uncompressed data is outweighed by the unavoidable JPEG artifacts on the output side.

So you are certain there's only "Yes" or "No", and there's no possibility of "Maybe"? What if the answer is inherently subjective and can't be answered definitively

Your question seems subjective to me. In my opinion, yes, there is a "yes" or "no" answer. The answer is "yes" if, on the whole, an end user will achieve superior results with Nokia's resampling compared to Photoshop in a variety of circumstances that is representative of all common shooting conditions. Otherwise the answer is "no". Even if Nokia's resampling only sometimes bests Photoshop, it would be acceptable to answer "yes" as long as it was qualified properly. Such an answer could state which resampling algorithm yields better results in which cases. For example, perhaps Nokia's resampling is really great in low-light shots and beats Photoshop, but in decent lighting there's no clear winner between them. If that was true, that's a major accomplishment that he should've described in his answer. But he didn't, which suggests to me that Nokia's resampling rarely, if ever, beats Photoshop by any substantial amount.

By the way, I'm still open to the possibility that Nokia's resizing could sometimes yields results that are superior to what a user can achieve by resizing the full-res image in Photoshop or other software. So far I've seen a lot of wishful theory but zero evidence. What really matters is the real world end result. Nobody seems to want to test this, though.
 

Bahamen

New member
Nov 20, 2012
272
0
0
Visit site
Guys, I know the full-res images are compressed. And I know there's a benefit to using the uncompressed data as input to the resampling. What I said is that since the phone only lets you save as JPEG, you - the end user - have to choose one of the following paths if you want the highest quality image at a particular size:

I really don't understand why, as an end user, that you need to choose any particular path. All you need to do, as an end user, is to select Pureview Mode, 2MP on the 808 camera itself. The phone will downsample from raw data file directly to the required size. In the example cited, 2MP. Nothing further needs to be done. Perhaps you are not familiar that the 808 allows downsampling to various sizes 8MP, 5MP and 2MP?

Your question seems subjective to me. In my opinion, yes, there is a "yes" or "no" answer. The answer is "yes" if, on the whole, an end user will achieve superior results with Nokia's resampling compared to Photoshop in a variety of circumstances that is representative of all common shooting conditions. Otherwise the answer is "no". Even if Nokia's resampling only sometimes bests Photoshop, it would be acceptable to answer "yes" as long as it was qualified properly. Such an answer could state which resampling algorithm yields better results in which cases. For example, perhaps Nokia's resampling is really great in low-light shots and beats Photoshop, but in decent lighting there's no clear winner between them. If that was true, that's a major accomplishment that he should've described in his answer. But he didn't, which suggests to me that Nokia's resampling rarely, if ever, beats Photoshop by any substantial amount.

Damian (or Nokia) WAS NOT making the case that Nokia's sampling beats Photoshop. And since they are selling cameras, not imaging software, I don't think he needs to go into a lengthy academic discussion on the advantages/disadvantages of downsampling versus Photoshopping. In fact, it is entirely possible that Nokia might not even have done any such comparisons with Photoshop! Seriously, are you now going to ask every camera manufacturer details on how their image processing algorithm compares to Photoshop? Anyway my main bone of contention is this: how are you going to prove any dishonesty or evasiveness. It is a serious accusation and I expect much more from you.
 

vlad0

New member
Oct 9, 2012
1,069
0
0
Visit site
both 1020&808 compress full resolution files, 808 even more(by default) since its 85 vs 95 in case with 1020

85% in auto/normal quality

95% in superfine

here is the difference:
85% jpeg compression

bfnorPM.jpg


This is 95%

PIFBrWK.jpg


A lot of online reviewers never change the 808's compression to 95% when comparing it to the 1020, which is at 95% by default.

I believe Damian meant that the 808's downsampling approach takes raw image data, before any JPEG compression, and uses that raw data for downsampling purposes. The alternative, as suggested by the interviewer, involves taking the 41/38/34MP JPEG (high res, but already compressed) file, and run it through Photoshop. Now, common sense dictates that downsampling directly from RAW data file is superior compared to downsampling from a high-res JPEG file. Whether or not this works well in practice is a separate matter, but Damian is certainly right that there are benefits to process an image directly from raw file as opposed to JPEG file.

Well summarized.

808 allows downsampling to various sizes 8MP, 5MP and 2MP?

I ma running a custom firmware that allows for even more options.. but I am not sure how effective those other resolutions are in terms of oversampling, its just nice to have them as an option, especially the 12Mpix one in 4:3

DVAHKfh.jpg


Wo80jBb.jpg
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
I really don't understand why, as an end user, that you need to choose any particular path. All you need to do, as an end user, is to select Pureview Mode, 2MP on the 808 camera itself. The phone will downsample from raw data file directly to the required size. In the example cited, 2MP. Nothing further needs to be done. Perhaps you are not familiar that the 808 allows downsampling to various sizes 8MP, 5MP and 2MP?

It seems like you may have missed the context of how this thread progressed. It's about whether there's any advantage in image quality for the end user whether they resize a full-res image down to the desired size using software such as Photoshop, or let the phone work its "PureView magic". You don't NEED to choose any particular path, or do anything, for that matter. But those are the options you have available to you, and one may yield a higher quality end result than the other. When someone claims that letting the phone do the resize yields higher quality, which is how this discussion began, I want to see the evidence. My tests and investigation have proven otherwise. And it's not about being anti-PureView or whatever. I just believe that the high resolution is the true key to PureView's breakthrough quality, not their downsampling techniques, and this is easily observable by comparing the two.

Damian (or Nokia) WAS NOT making the case that Nokia's sampling beats Photoshop. And since they are selling cameras, not imaging software, I don't think he needs to go into a lengthy academic discussion on the advantages/disadvantages of downsampling versus Photoshopping. In fact, it is entirely possible that Nokia might not even have done any such comparisons with Photoshop! Seriously, are you now going to ask every camera manufacturer details on how their image processing algorithm compares to Photoshop? Anyway my main bone of contention is this: how are you going to prove any dishonesty or evasiveness. It is a serious accusation and I expect much more from you.

Again, context of the thread. I never said Nokia made any claims to that effect. In fact, I even explicitly said they DID NOT make such claims, earlier in this thread! So, when a tech journalist asks that sort of direct question, I think it's reasonable to expect a more direct answer. He was frank in most of his other answers in the interview. And we're not just talking about Photoshop. The journalist clearly used that as an example of the more broad class of standard image resizing algorithms that people rely on every day in the real world. I think given all the technical explanations and promotion they did of the downsampling, they should have a pretty good idea of where the benefit is really coming from. The heart of this question is really: how much of the success of PureView is attributable to the high resolution, and how much is attributable to the downsampling? I stand by my assessment of his answer as evasive. I'll give him the benefit of the doubt and assume he wasn't being dishonest - although I still find it extremely hard to believe they wouldn't have made such comparisons. But his answer did come across (to me) as evasive, even if that wasn't his intent.

In any case, if Nokia's downsampling did offer any significant benefit over standard resizing, this was a great opportunity for him to tout that. He didn't. I think his last sentence sums it up well: "a slightly different balance"
 

vlad0

New member
Oct 9, 2012
1,069
0
0
Visit site
I just believe that the high resolution is the true key to PureView's breakthrough quality, not their downsampling techniques, and this is easily observable by comparing the two.

Well.. if you get down to it its the size of the sensor and amazing optics from CZ.

There is an ongoing debate on pixel size vs. pixel density , but in Nokia's case .. they had to go with pixel density instead of physically bigger pixels since that would require bulkier optics.

more here

Pixel density revisited: News & Rumors Talk Forum: Digital Photography Review

The Joy of Pixel Density: Open Talk Forum: Digital Photography Review

So.. I think that the size of the sensor is the biggest contributor to what the 808 can achieve.

I think that its very interesting to compare the Sony RX100 and the 808 since they those two different paths, which I mentioned here:

http://forums.windowscentral.com/no...ing-808-symbian-lumia-1020-a.html#post2163068
 

vlad0

New member
Oct 9, 2012
1,069
0
0
Visit site
^ yes, especially for video.. love the crop/zoom :)

But overall, if you look at the sensor sizes of the "top" camera phones out there.. the IQ level pretty much corresponds to the size of their sensor

the bigger the sensor the better the IQ

md6SPej.jpg
 

Nick_1020

New member
Mar 4, 2013
1,014
0
0
Visit site
I haven't read the last few pages of toing and froing on this issue but I will say this:

As much as I love the 1020 as a smartphone, I think the camera needs quite a bit of tuning. Whether that is possible remains to be seen but I'm hoping that it is possible as I'm beginning to look on my 808 images with a new found fondness.
 
Last edited:

Bahamen

New member
Nov 20, 2012
272
0
0
Visit site
It seems like you may have missed the context of how this thread progressed. It's about whether there's any advantage in image quality for the end user whether they resize a full-res image down to the desired size using software such as Photoshop, or let the phone work its "PureView magic". You don't NEED to choose any particular path, or do anything, for that matter. But those are the options you have available to you, and one may yield a higher quality end result than the other. When someone claims that letting the phone do the resize yields higher quality, which is how this discussion began, I want to see the evidence. My tests and investigation have proven otherwise. And it's not about being anti-PureView or whatever. I just believe that the high resolution is the true key to PureView's breakthrough quality, not their downsampling techniques, and this is easily observable by comparing the two.

I've said it earlier, and I will say it again. The Pureview's algorithm has the advantage of working directly with raw image file. Any third party processing solution requires working from a compressed JPEG file. Without knowing anything about the algorithms used (whether Pureview's is better than Photoshop or not), it is certainly a fact that working from a raw file is better than from an already compressed JPEG. And that is exactly what Damian said "you're probably better off doing it at point of capture."
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
I've said it earlier, and I will say it again. The Pureview's algorithm has the advantage of working directly with raw image file. Any third party processing solution requires working from a compressed JPEG file. Without knowing anything about the algorithms used (whether Pureview's is better than Photoshop or not), it is certainly a fact that working from a raw file is better than from an already compressed JPEG. And that is exactly what Damian said "you're probably better off doing it at point of capture."

Yes, but what good is that advantage when you're resizing the image down substantially in both cases anyway? The JPEG artifacts in the compressed full-res image get killed by the resize, just like noise and other undesirable crap. I bet even under very close examination, you wouldn't be able to tell the difference between a raw 38MP image resized to 8MP vs. the 1020's JPEG-compressed 38MP image resized to 8MP.

However, I guarantee that you will be able to see the JPEG artifacts when the resized image has been encoded as JPEG. Possibly without even needing close examination depending on the image. I showed this in the images I posted earlier in this thread. If you let the phone do the resizing, you're forced into having these artifacts. If you do the resizing yourself, you can get a totally clean result. It's up to you whether you want to encode the result as JPEG and at what quality level. And it's a better base for further processing, without working on top of an already JPEG artifacted image.
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
Well.. if you get down to it its the size of the sensor and amazing optics from CZ.

There is an ongoing debate on pixel size vs. pixel density , but in Nokia's case .. they had to go with pixel density instead of physically bigger pixels since that would require bulkier optics.

I look at it in a simpler way. If the camera has a high resolution, and it yields a half decent result at that full resolution, you're going to be in good shape if you ultimately want images that are substantially lower resolution than that. So the 808 and 1020 do a pretty decent job at 38MP. That means when you want something like an 8MP image, you can get a great result! Now, if someone came along with a camera whose max resolution was 8MP, but it looked as good as the 1020's 38MP image resized down to 8MP, then it would be a different story. But so far I don't think that's happened, right? I think even DSLRs don't look so pixel-perfect at their max resolution. You always need to resize down to get that pixel-perfect look. That's why it's so important to have high resolution... in today's world.
 

vlad0

New member
Oct 9, 2012
1,069
0
0
Visit site
Now, if someone came along with a camera whose max resolution was 8MP, but it looked as good as the 1020's 38MP image resized down to 8MP, then it would be a different story. But so far I don't think that's happened, right?

That can easily happen if the sensor on the 8Mpix device is big enough. In fact, you will get better dynamic range because of the bigger pixels which would results from that.

Look at it this way, take the 1020 sensor and cut it into 8Mpix instead of 41Mpix you will get the following:

The sensor size is: 8.80 x 6.60 millimeters

8Mpix image = 3248 x 2448 depending on aspect ratio

In 1 mm we have 1000 microns

8.80 x 1000 = 8800 microns across

8800 / 3248 = 2.70 microns per pixel

So you go from 1.1 to 2.7 producing an 8Mpix image.

I am willing to bet that, everything else being equal, the 8Mpix image from the bigger pixles will actually be better than the software one we get from the 808 and the 1020.

Here is a cool graph showing the different size of the pixels at different resolutions, it should be similar for the 1020.. slightly smaller.

9Lgc1sX.jpg


So like I said.. there is an ongoing debate on pixel density vs pixel sizes , and it seems like density might be the way to go after what Nikon did with their D800..

. Without knowing anything about the algorithms used (whether Pureview's is better than Photoshop or not)"

Per this wonderful French review:

Test du Nokia 808 PureView : Photos, Vid?os | Test-Mobile.fr

They used something very similar to the Lanczos resampling algorithm... which you can manually apply to a processed jpeg, but from what I've seen, the on board engine tends to remove noise in a more effective manner, that's why I shoot my low light images in pureview mode instead of full res and resize later.


I haven't read the last few pages of toing and froing on this issue but I will say this:

As much as I love the 1020 as a smartphone, I think the camera needs quite a bit of tuning. Whether that is possible remains to be seen but I'm hoping that it is possible as I'm beginning to look on my 808 images with a new found fondness.

They might be able to get it closer, but I don't think that they will ever be able to match the 808 with the available hardware on the 1020. And yes, once you get used to those clean 808 jpegs, its very hard to accept anything else after that.

Personally I think everything comes down to this

sensor-size-comparison-lumia-1020-808-pureview.png


Of course, BSI has some advantages in low light, but during the day.. a smaller sensor is a smaller sensor
 
Last edited:

Tech friend

New member
Sep 26, 2013
73
0
0
Visit site
I'm beginning to look on my 808 images with a new found fondness.
And you are quite right to do so. :smile:

I think the camera needs quite a bit of tuning. Whether that is possible remains to be seen.
I find that uncertainty to what extent Nokia will (be able to) improve the camera software quite uncomfortable. Besides their announcement to look into the white balance issue I didn't found much more hints what they are going to do in order to enhance the camera.
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
Per this wonderful French review:

Test du Nokia 808 PureView : Photos, Vid?os | Test-Mobile.fr

They used something very similar to the Lanczos resampling algorithm... which you can manually apply to a processed jpeg, but from what I've seen, the on board engine tends to remove noise in a more effective manner, that's why I shoot my low light images in pureview mode instead of full res and resize later.

Good find, thank you. There's clearly chromatic noise present in the dark blue sky in the resized full-res image. Assuming this is not related to JPEG compression (which it shouldn't be), there appears to be a slight advantage to the PureView modes in low light. I wonder how the 1020 would do with this.
 

Bahamen

New member
Nov 20, 2012
272
0
0
Visit site
I never said Nokia made any claims to that effect. In fact, I even explicitly said they DID NOT make such claims, earlier in this thread! So, when a tech journalist asks that sort of direct question, I think it's reasonable to expect a more direct answer. He was frank in most of his other answers in the interview. And we're not just talking about Photoshop. The journalist clearly used that as an example of the more broad class of standard image resizing algorithms that people rely on every day in the real world. I think given all the technical explanations and promotion they did of the downsampling, they should have a pretty good idea of where the benefit is really coming from. The heart of this question is really: how much of the success of PureView is attributable to the high resolution, and how much is attributable to the downsampling? I stand by my assessment of his answer as evasive. I'll give him the benefit of the doubt and assume he wasn't being dishonest - although I still find it extremely hard to believe they wouldn't have made such comparisons. But his answer did come across (to me) as evasive, even if that wasn't his intent.

In any case, if Nokia's downsampling did offer any significant benefit over standard resizing, this was a great opportunity for him to tout that. He didn't. I think his last sentence sums it up well: "a slightly different balance"

Even if Nokia's downsampling were mere simple resizing (albeit with a different implementation), I really don't see a problem with that. Nokia never claimed that their algorithm is inherently superior compared to other resizing methods, they only claimed that downsampling helps with image quality, which is true. Due to the different implementation, it may give different results than if you used standard methods, which is again true. It is sampled directly from RAW which may possibly be better, which is, again, true.

The heart of this question is really: how much of the success of PureView is attributable to the high resolution, and how much is attributable to the downsampling?

No, that's not the heart of the question. The heart of the question was that posed by the GSMArena interviewer whether Nokia's downsampling is superior compared to Photoshop or other image editing software, Damian replied they haven't really made that comparison, and you basically accused him of lying.

Look, I will just cut to the chase here. My only problem here is with your attempt to cast aspersions on someone's character without any concrete evidence when you are only making fast and loose assumptions. We can have a debate about whether Nokia should have done a comparison with Photoshop. But if Damian said they never really did such comparison, and there could be many explanations for that (one being that it is not necessary for them to do so), "he's lying" should not be the default explanation. I don't know about you, but I tend to take these things seriously when you start to publicly question someone's honesty and integrity.
 

tgr42

New member
Jul 31, 2012
286
0
0
Visit site
Look man, I never said he was lying - that was all you. Stop putting words into my mouth. I said his answer was evasive - which is an extremely common way of responding to difficult interview questions - and I stand by that assessment. If you don't agree, fine, it's an opinion.
 

Bahamen

New member
Nov 20, 2012
272
0
0
Visit site
Look man, I never said he was lying - that was all you. Stop putting words into my mouth. I said his answer was evasive - which is an extremely common way of responding to difficult interview questions - and I stand by that assessment. If you don't agree, fine, it's an opinion.

Really? What about that quote below? And why does the situation even warrant your generosity in granting him the benefit of doubt that he is not lying?

He was caught in a difficult question and did his best to weasel his way out of it. I do give him credit for attempting to give a "least untruthful" answer, although his remark "we haven't really made that comparison" is pretty shaky.

Weasel? "least untruthful" not the same as lying?
 

Members online

Forum statistics

Threads
322,913
Messages
2,242,887
Members
428,005
Latest member
COME ON WIN ANDROID (ADI)