Digilux 2 / RAW question -

[Lonestarchitect]

Advertisement (gone after registration)

One more question regarding this camera, and I promise to go away for a bit

 

Here is my problem: The D2 is a 5MP camera. For work, I need images that have 3100 pixels on the long side (basically an 7.2MP image). Lower than this, and I don't get paid. So, I like to get paid, but I also want to keep using the D2 (for the same reasons a majority of the people on this forum like using the D2). I'm not going to double shoot everything.

 

So, upsampling a .jpg doesn't give good enough results, but I was wondering if anyone had tried upsampling a RAW image . . . . ?

 

I don't normally shoot in RAW (again, for the same reasons most people don't shoot in RAW with the D2), but in my mind, I think upsampling a RAW image might.

 

My rationale: the D2 has a full size image sensor, 1.6 times larger to be precise. 5MP x 1.6 = 8 MP

 

Life is rarely this simple, but I think this is why this camera is sometimes compared to an 8MP equivalent. Would this work? I would just try it, but I really just don't know enough about RAW photo editors.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[audidudi]

There's really only one way to tell: Try upsampling some RAW images and see what you think. Personally, I think the software included with the D2 sucks and I was (am) happy enough using the free version of Rawshooter to convert the camera's RAW files into .tif files instead before post-processing them in PhotoShop (although if you're willing to spend the money -- this is a profession for you, after all -- do be aware there are better, more versatile programs available.)

 

That said, it's been my opinion that as wonderful as the D2's images look on my monitor, the upper size limit for prints for me (YMMV) is about 6" x 8" and sometimes 7.5" x 10", depending upon the subject matter and regardless of what sort of post-processing I do. For the images I shoot with this camera, that's generally good enough and Yes, for the record, I'm probably a bit pickier than most about this sort of stuff, so I have no doubt others will tell you they can print larger still with good results. (FYI, for architectural work, my preferred setup is my 8x10 Toyo view camera loaded with Fuji Astia or Provia, so you can see where I'm coming from as regards resolution and print size.)

 

But if I were you and I absolutely must have images that are a minimum of 3100 pixels across the long side in order to get paid, then I would look for another camera. In this situation, the D2 is merely a tool and trying to achieve professional results while using the wrong tool for the job makes no sense to me. If you like the basic look of the D2's images and want a camera with a similar operational feel, check out the D3 ... it'll give you your 3100 pixels with a few to spare (its sensor is 7.5MP to the D2's 5MP) and is very flexible with regard to your choice of lenses. I have an L1 (the Panasonic version of the D3) and use it happily when I need a file larger than the D2 can create or to use a lens outside the D2's range.

[Lonestarchitect]

I should clarify my needs a bit. While I do get paid for taking these photos, it is incidental to my work, a bonus, more or less. And the D3 just doesn't feel right in my hands . . . so if it doesn't work, I'm still going to shoot with the D2, and just not be able to sell these photos.

 

I did do a trial run (I didn't know it, but apparently Photoshop handles RAW - shows how much I know about RAW photography.) Honestly, a RAW image that has been upsampled looks better to me than a .JPG that has been upsampled, but without having a 7.2 MP 'native' camera shot to compare it to, I'm not really comparing apples to apples. My gut says that the upsampled RAW image is still better than an image taken by a inexpensive 7.2MP camera, but I have nothing to compare it to.

 

I guess my question might be better approached from the 'theory' side of this, ie, does upsampling in RAW effectively 'convert' the larger image sensor in the D2 to a smaller image sensor, and produce an image comparable to a 'standard' 7.2 MP camera? This might be a question for one of those people who like to explain why a 35mm lens on a film camera does not behave the same way on a digital camera . . . .

 

The funny thing about this is that I could take a picture with my Sony DSC-T77 (which has never and will never take a good photo, since the lens is about 4mm across) and that would be OK.

[laptoprob]

What happens in upsampling? The sofrware has to invent extra pixels between the original ones. If you would do a 2x upsampling, there is one new pixel between every original two.

So how does 1,6x upsampling work? It can only mean every single pixel is re-invented. Every two pixels have to become about three. Can't be what you need.

[ho_co]

LSA--

When Leica introduced the Digilux 2 in Houston, they showed a very nice 30" x 40" portrait made from it.

 

There are a number of heavy-duty upsampling programs on the market; one from Alien Skin has been highly recommended on the forum.

 

But late versions of Photoshop also have good upsampling capabilities as well. There was an article on the topic in the last six months or so by Ctein on The Online Photographer.

 

RAW is well worth the learning curve, and up-resing the files from the Digilux 2 such a small amount should work just fine, even though it's another step in the workflow.

[Arman]

This is an interesting question.

If I understand you correctly your question is whether you could create an upsampled image from a Digilux 2 which is at least as good as a normal (I mean mainstream) 7.2 Mp digital camera?

 

I upsampled a RAW image from 2559x1919 to a 3100x2325 image right away and the overall quality didn't become visibly worse. Probably the contours of fine details will blur a little bit but I think it is ofcourse better to wait with the "sharpen" filter untill after the upsampling.

 

But it should be possible to optimise this operation by choosing the right (or rather "best") way to resample the image (e.g. bicubic, bilinear, etc)

 

Regards

Arman

[ravduc]

Advertisement (gone after registration)

You will not notice any difference whatsoever between a 5 and a 8 megapixel camera with the same size sensor on a print, even an 11x14 inch print. The difference that a previous member mentioned between the Digilux 2 and 3 (or L1) is most likely due to the difference in sensor size. The sensor in the D3 is approximately twice as large as in the D2. As for upscaling your prints, if you do not exagerate, then you will not notice any difference on a reasonable size print. I have upscaled 5 megapixels to at least 8 without any obvious loss of quality up to an 11x14 print. I think that you need to try this out and compare this to other higher pixel cameras with the same size sensor. I have done this mainly because I happen to have several digital cameras to compare (Olympus E1, E3, E300, Digilux 2 and 3).

[John Thawley]

Your 5mp file from the Digilux 2 should have no problem spitting out a 16x20 print. Most commercial labs don't require any more than 144 pixels per inch. So with a little upscaling, you shouldn't see any loss in image quality.

 

If you're going with fine-art prints, sure... you need to revisit your methods of interpolating the file upwards... but it's not that big of a deal.

 

JT

[ho_co]

My rationale: the D2 has a full size image sensor, 1.6 times larger to be precise.

Lone--

Not that it makes any difference, but I'm unsure what "full size sensor" means.

 

D-Lux 4's sensor has 9.2 mm diagonal (by my calculation).

Digilux 2's sensor has 11 mm diagonal.

FourThirds sensors have 22.5 mm diagonal.

M8's sensor has 32.4 mm diagonal (3:2).

"Full Frame" has 43.3 mm diagonal (3:2).

 

But whatever those numbers indicate, the images from the Digilux 2 can easily be up-resed to the size you want.

[Lonestarchitect]

Thanks for all the good responses - there is a lot of aggregate knowledge on this board! Also, thanks for the suggested reading material. It was very informative.

 

In order to meet the VRA (Visual Resources Association) requirements for archival, I only need to upsample by 1.21x or so to get to my target, so it looks like it should be fine - I'll do a few more tests and try a few of the plug-ins (thanks for the heads-up on the Alienware plug-in - they are talking about 'serious' upsampling, starting at 150%, so I don't feel like I'm really changing much. Plus, it looks like they've just released a new version!), and in particular I'll contrast the differences between RAW upsampling vs. JPG upsampling, and see if it makes any difference . . . . and I hope it doesn't, because I'd rather shoot JPG

 

Thanks again,

 

-sam

[Overgaard]

For whatever it's worth, is upsample straight to the desired size and never had any problems. I do so from JPG and some times, when required, I have to change the format into TIFF for ridiculous reasons because some people sit around and still think that the bigger a file, the better quality.

 

But in fact, if you look at a file, that's what it looks like. Changing size will just see to that it doesn't pixelate in print. That's all.

 

A wool-looking shot in 5000 pixel wide hires doesn't stand a chance faced with a sharp upscaled D2 JPG to the same size.

 

One could say that A4 (7x12 inch) 300 dpi resolution is enough for most of what you can think of using if for. It's enough for a full page magazine print, it's enough for full-size newspaper print (as they require only 150 dpi or less why you can go larger with the same file). Few require a larger than that print in any quality that require 300 dpi 1:1, and then it would anyways require medium format to get it. Most poster will be less dpi and viewed from a distance.

 

Not that I support lowering quality, but don't panic if you have a great shot that needs upscaling.