Friday 1 February 2019

S3is - Postscript: raw 2019

Looking through my S3is images left me curious about one thing -  I wondered whether, using my current post processing techniques I could produce images 1400 pixels high of acceptable quality for viewing on screen. It isn't that I want to use the S3is now. I have cameras with much better ergonomics, larger screens, more customisable, better image stabilisation, better sensors etc. But on the photography forums I frequent there are, unsurprisingly, a lot of discussions about kit, what kit you need to get good image quality for various types of subject matter. One of the criteria I see quite often is along the lines of having to have good enough kit to produce prints a metre or more wide. That is far beyond my needs. I don't print much and when I do it is rarely more than 300 mm on the long side. And the vast majority of my photos are prepared for viewing on screen at 1400 pixels high, on a calibrated monitor, in subdued light.

Why 1400 pixels? It fits nicely on to my current screen. Perhaps it will be too small when I get a 4K screen but I'll worry about that when it happens. But why not have a lot more pixels so people can blow it up and look at the fine details? Two reasons. The way I go about my photography, especially with my small sensor cameras, there often isn't much detail worth looking at beyond what you can see taking a close look at 1400 pixels high. And, even if there is, I want to be in control of the look of the image, and that includes the sharpening. For my invertebrates especially, and sometimes for flowers too, fine detail matters. If an image is resized it can destroy the fine detail, so I prepare my images, and do the sharpening, for them to be best viewed as is, at the size they are posted. Of course if someone doesn't have a screen as large as mine (2560 x 1440) then unfortunately they will have to look at a resized and hence degraded version if they want to see the whole image. That is a pity, but you have to draw the line somewhere, and the line I draw is that it looks good on my screen. Selfish? I suppose so. But you have to choose a particular size, so it is always going to be wrong for some people, so I might as well choose the size that works best for me.

And the calibrated monitor, and subdued light? One of the things I discovered when I started using a calibrated monitor was that I could see more, finer detail, more subtle textures and finer colour gradation than before, and textures and colours are very important for botanical images for my viewing enjoyment. And viewing in subdued light helps with this too.

So, I want to produce images 1400 pixels high. How good a piece of kit do I need to do this? Could I do it to my satisfaction with this old camera with only 6 megapixels and a sensor the size of a smart phone sensor, with a bit of cropping along the way? I thought it quite possible, because my go to camera for invertebrates is a small sensor camera (a Panasonic FZ330) with a sensor only fractionally larger than that of the S3is, and I sometimes crop quite hard and produce images 1400 pixels high. But it does have twice as many pixels as the S3is and a more recent and so presumably better sensor. And I shoot raw and use some heavyweight processing, using DXO PhotoLab, then Silkypix Developer Pro and then Lightroom for invertebrates, using the best parts of each (for my purposes and preferences).

I decided to do a practical test. This is one of the common features of my photographic journey. Right from the start I have been doing practical exercises to try to understand how things work, and how well they would work, or not, for what I want to achieve, and sometimes just out of interest, as with this one. Most of the images on my Flickr account are arranged by year, and each year I have a collection called "Photo Technics", with an average of around 75 albums per year in them over the past four years (13 in January alone this year). This is where I put the photographic evidence from my practical tests. And these tests are why my Flickr photostream can look so odd sometimes with large blocks of almost identical images. A much better entry point for my Flickr account is my collections.

A good test would have been photographing invertebrates, but there are none around at the moment that I can find. (Much of the country is covered in snow just at the moment.) However, I have been doing exercises with botanical subjects in our garden over the past month and so have been looking closely at images of the same dozen or so subjects, so I would have a decent feel for how well or otherwise the S3is did with those same subjects.

So, I fitted a Canon 500D on to the S3is and spent a short time in a rather cold garden, working hand-held as usual, shooting some familiar botanical scenes. One thing I did first was to install the CHDK utility on the S3is so it could capture raw as well as JPEG. There are periodic discussions on the forums as to whether there is any benefit in using raw rather than JPEG, especially when using small sensor cameras. My thinking for some time has been that there is a benefit for what I do, although it does depend on the techniques I use. We shall see a lot later in the story why it is that I am now using mainly JPEG for my botanical imaging, but that is a long way ahead. Suffice it to say that I have an open mind about raw vs JPEG; raw works for me, for some of what I do, but it doesn't work for some other people, for anything that they do. In this case I decided to shoot raw + JPEG and compare the results.

The JPEG settings were those I used back in 2007-9, with contrast, Sharpness, Saturation and Green turned down as described in this post, so as to improve the colours and give some extra flexibility for processing the out of the camera JPEGs. While capturing the images I used manual white balance and set it from time to time using the large grey patch on a ColorChecker Passport. I used aperture priority with exposure compensation of -2/3 stop to protect the colours in brighter areas.

The first thing I discovered when I got back indoors and loaded the images on to my PC was that DXO PhotoLab does not know about the S3is. That meant I could not use PRIME noise reduction. That was a blow. Small sensors can be a bit noisy even at base ISO in good light. I typically underexpose my images, as I did for this exercise, to protect highlights, and I go in for a fair amount of shadow lifting. This reveals noise, which is why the excellent PRIME noise reduction is especially useful with my small sensor cameras. Also, the noise gets worse as the ISO increases, and I almost always use my FZ330 at base ISO 100. I can do this because I use flash for invertebrates. However, I didn't have the S3is set up for flash, and in any case I don't use flash for botanical subjects. But unfortunately the light levels were fairly low so even though I was using f/4 rather than the more light-demanding f/8 that I use for invertebrates, I had to use ISO 200, and even then the shutter speeds came out at no faster than 1/160 sec and down to 1/40 sec, which is getting a bit slow for hand-held close-ups. At least there was not much of a breeze.

I couldn't use PhotoLab on the S3is raw files, and I don't use Silkypix for botanical subjects, so I did all the processing in Lightroom. Now things get personal. What one can get out of an image, raw or JPEG, depends on the tools one has available and one's skills and experience in using them. And what one wants to get out of an image is a very personal affair, as is one's perception of and tolerance for things like noise, wrong colours and sharpening halos. So I can only describe how it felt to me.

For the most part, compared to processing an out of the camera JPEG image, the raw files seemed easier to get to a place that I found more or less satisfactory. The colours seemed to fall into place quite easily. Some colours in some of the JPEGs, seemed stubbornly not quite right despite my attempts to adjust them with white balance and individual colour adjustments. I think on balance the colours of the raw versions turned out more like the real thing, but it wasn't clear cut.

Another thing that was difficult to pin down but which felt real enough to me was that the raw versions seemed to have more "dimensionality" than the JPEG versions, which seemed a bit flat in comparison. I suppose this would have been (assuming it wasn't my imagination or eyesight) something to do with overall contrast and/or local contrast, or the tone curve. (I tried altering the tone curve on the JPEGs but I didn't get anywhere with that.) Quite possibly with better post processing I could have got the JPEG versions to be more like the raw versions that I preferred, however I didn't need to do anything special to get the effect with the raw versions and that in itself was advantageous.

There was one thing which was clear cut. In the following JPEG version of one of the scenes there is an area in the bottom left that is I believe irretrievably blown.


The same area in the raw version below is fine.


So was the S3is able to produce 1400 pixel high images that I would be comfortable with? Well, the results varied, which is hardly surprising given the shooting conditions, but I think most of them would be just about good enough for my purposes, especially if did some local noise adjustment on some of the backgrounds (something that I have to do from time to time with FZ330 images). Here, below, are the rest of the raw versions, 1400 pixels high. (If you would like to look at the JPEG versions they are paired with the raw versions in this album at Flickr.)

In terms of this test the landscape orientation images are the most challenging, as the vertical dimension is at most 2112 pixels. The one that was most cropped in this dimension was 1817 pixels high when cropped, with two others being in the mid 1800s. In the context of a 1400 pixel high output I think this small sensor camera has probably done about as well as could be expected. Perhaps a bit better than some might expect.

Click on an image to see the 1400 pixel high version


















No comments:

Post a Comment