Søk i denne bloggen og Lenker

onsdag 21. september 2011

BW Grain Experiments in GIMP for JPEG Images From the E450 Olympus DSLR

Black and white images out of the camera can be dissappointing, and even more so when they are "monotonised" or de-saturated from colour shots. There are several very comprehensive bloggs and chapters in books about getting effects which mimmick some of the classic photo journalist and artistic shots from the hey days of ilford fp5!

In this blogg I will offer a simplified, very quick approach to adding a corn or heavy grain effect to images. In the days of film, the grain was often caused by photographers "pushing" the negatives or prints to make a dark image captured at silly shutter speed, light enough to get in print. Alternatively, the enlargements were of a small area of the negative, or they used a small negative in the first place, like the original PEN crop onto 35mm.

Firstly you will need a texture background with an element of random grain and you can create these yourself at vast hassle or get one like the png file from http:byscuits.com ( it is a big file and will slow your browser down) grain-400tm.png. Dave Gandy has hosted this big png on behalf of the owner, and has this as a kind of traffic dirivng advertisement for visitors. Thank you both Petteri S, and Dave G . for allowing this to be shared.

This particular image has some "Hairs" on it and is actually quite a fine grain. It is like an ISO 400 film I beleive, and at that level of "push" in BW you don't get the real effect- a half way house between sharp, natural textrure and impressionable grain.

1) Open and Adjust the Grain Image to Corny-ness, size and Orientation.

So the thing to do is then open this in GIMP and then clip out a section to be rescaled and therefore have more visible corn. The grain 400 png is about a 20mpx scan, landscape so you need to go to a small crop in order to get a heavy grain effect for a 10mpx image in the case of the E450.

Open as the first image, clip to between 10 and 25% of the area with the tool to either portrait or landscape rough dimensions. Then use in the image pull down ( scaling layers does not work to scale the grain once you have two open). To super impose this grain on a landscape it is W3648 x2736 for the E450 at 10mpx jpeg, and conversely 2736 x 3648 portrait 90'.

At this point it is worth saving the grain crop scaled to 10mpx, and probably worth making a set of different grain densities in portrait or landscape orientation. PNG is fine, but an xcf layer ( gimp's own bitmap with extended information on process) is as good or it can be okay as a jpeg.

2) Open The Image You Want Grain / BW onto as a Layer

Gimp seems to detect that the png layer is monotone and converts the incoming jpeg from colour to BW automatically: but if it does not in your case / version /universe then you can just select that layer alone and choose - desaturate - from the colour pull down.

If it is not open already, use CTRL-L to open the layers dialogue box.

3) Adjust your image layer tonal curve

If you are au fais with using the tonal curve adjustment then you should adjust the contrast upwards in that way, and also see if there is some quality of the image to be gained in the individual R-G_B channels. If you are a simple mortal just into GIMP and no more, then use contrast slider until you like the image.

The image will lose a bit of contrast and brightens when the grain layer is merged in, so you need to experiment and make sure you are looking at the layers properly: image on top of grain background, both "eyes" on in the dialogue showing they are both visible when you want to see the combined effect, and only the image is visible when you adjust that alone.

3) Adjust opaicity or use - Grain Merge -

This is the reverse of transparency. 95% is then only 5% see-through. Adjust this using first the "Normal" selection in the mini pull down on the Layers Dialogue window.

As stated you will lose some contrast which you then have to compensate for, and you may eventually need to darken or lighten the grain layer in order to get the desired texture.

You can then experiment with all this, but GIMP has done much of the job for you in mergign texture without altering the contrast too much, employed via the same pull down, quite far down the list you will find " Grain Merge" and this instantly shows the merging of the texture to the image : it still does change the contrast, adding some white and grey quarter tone noise from the png file: but you can then adjust the curve of the image more subtely and see the result directly

3) Save An "xcf" File " Save a copy..."

when you are near finished, do this: it is worth doing underway too, so you can trace your work back. FOr example if it was the grain layer which really messed up the contrast

4) final adjustments.

Fine adjustments to the curves, any sharpening you feel you may want ( with threshold above 8 pixels so the grain is not "worked up"). In the byscuits grain png there is are abberations, notably hair lines which have white trails left on the image at some point in its processing.

5) Merge down and Save

Merge down is the final thing you need to do: when using GRAIN MERGE you get a perfect preview of the merge, while with semi opaque you can have an issue sometimes with final versus your impression from viewing the layers.

If you choose to save a copy now as a jpeg, then it will merge and export to jpeg., leaving you free to save as an xcf and also work backwards in "undo-s" if when in the cold light of day you hate it all!!

fredag 9. september 2011

Extended dynamic range using Layers from Bracketed Exposures 1.

One criticism of the 43 sensor is that it has limited dynamic range compared to mid to high end APS-C sensors. This means that highlights become white blocky areas and lose realism, and shadows the same in dark slabs. In fact the criticism is that the top of quarter tones, light with detail, and the bottom of three quarter tones actually disappear into highlight and shadow in an otherwise well exposed shot, leaving an area which is unrealistic, in being blocky, posterised or marked by noisey speckles.

In a perfect world we would have sony's sensor qualities in the D7000 in four thirds, and there will be progress in this at least within 12-14 megapixels according to many enthusiasts who know a thing or two about quantum physics.

However we want to acheive what we can within our limits or as part of the challenge of mastering our equipment and post processing (post') technique.

A simple means to capture more dynamic range is to make bracketed exposures, and then layer them onto each other. The top layer or even several layers are made partially transparent.

The best technique is to use a tripod and take three immediate shots, especially in daylight so conditions and shadows will be as near as possible between each bracketed shot. Most all enthusiasts cameras bracket on increments (+/-0.7 is a good starting point, giving three bracketed shots). Otherwise cameras often have an exposure adjust, which allows for slight under and slight over exposure to +/- one exposure step. A simpler but less effective means of doing this is without bracket function is to set the camera to aperture priority and simply take one shot "right" and two other shots either side of that. Alternatively you can as I have below, set up manual exposures which vary the depth of field as well as the overall exposure.

From the cameras own LCD /LED screen it can be difficult to judge the exposure of each shot, but you can see that some are quite dark while others become too bright. You can look at the histogram to see if there is a large skew to the last 5% of either side of the scale. However some overly dark shots on screen actually have a lot of detail to be tweaked back out of them.

So I recommend setting up with a tripod and running lots of shots: get a feel for the depth of field ( we are going to consider a long depth of field which gives a sense of "realism" in combination to extended dynamic range). The second advantage of a tripod is you can shoot unusually long exposures in daylight @f18-22 or in addition an ND or polarising filter.

Example Using Gimp

This is my next experiment, having come a long way in adjusting jpegs in curve work, sharpness and so on. It helps if you have worked in layers before, but in fact this can be the simplest use of layers (as apposed to the long route to masking to blur a background in an earlier blog).

For my first ever venture into layering for extended dynamic range, I chose some night shots with very long exposures. I wasn't happy with the end results in either of two exposures : one was too dark and the other was light and soft with flat highlights around some nice period style lamps.

I wanted to try several exposures but on the laptop I wasn't pleased with either of the two "keepers" from the back of the camera view. However both had their merits. So my first fodder for a long awaited experiment.

So what came out of the box??

One too dark: maybe a lift would help it, but then it may lose depth and realism. The other one is about right, but a bit too soft and short a depth of field. Also the highlights are overcooked around the detail lamps I wanted.

Try One: Dark onto of lighter: shining through?

Okay, so these have uploaded not so great but you can see there is a subtle extension to the DR in the merged down shot, dark ontop of light with about the right transparency (sorry I forget if it is 60% + or under 30%: experiment yourselves, use undo to get back !)

The highlights are still blown and the detail is not sharp. A kind of worst of both worlds but the exposure is better.

Now the reverse: a thin, 24% light layer ontop of a 65:8:5 sharpened dark layer which was lifted slightly in the shadows. The light layer on top is slightly dropped on the highlight peak (lamps of course)

Now! Here I have had quite a plasticy looking "dark" layer once sharpened, but the feint light layer is merged down at 24% opaicity it unexpectedly softens it nicely. The main aim was to lift the shadows.

Any benefit over just one image worked up on curve and sharpening though?? Well as mentioned the sharpening which hardened the details on the lamps, balcony, woodwork etc would be too intrusive while the image with the best DOF viz the darker, if chosen would become noisy and dull when lifted so much!!

I have heard of layers being used from negatives in the dark room, suppositioning several exposures onto one paper. And of course it is the origin of the colour process!

So I am pretty pleased and once again I feel a little of the magic that people had 30 years ago and longer when an image would appear out the baths magically.

Here is the image larger below because I can see the compression is killing the IQ : however you can see what I mean above in all the shots.

fredag 2. september 2011

Mu Ft y

Well I am disappointed that sony is going over the top in spec and price for the NEX 7 while Nikon look like following Pentax into an early grave with limited, price point competing interchangeable lens cameras.

What do I like about mFT?

Same things I like about FT- size, entusiast spec in controls and menus, expandability

Then there is : expandability..in 2020....the range of bodies, the new choice

-ve? CDAF ?? the range of new cameras...will I miss out on IR laser range finding which will be PDAF quality if i buy the flavour of next week? or built in EVF?

Sensor quality gap: will it be worth it? I mean a d7000 for 'serious' thought out shots and a ZX1 for walk around shots capable of wowing or sligjhtly impressing people on Facebook?

PRICE of decent kit. The price per f stop has to come down!

Unfortunetly Nikon isn't going APSC or a crop onto their 16mpx chips set up to say 12. They are allegedly doing a small sensor and a slow mega zoom according to the rumour mill. Bummer . Panolyeiaca will feel no pressure to do anything more than the EPM 3 and maybe an entry level from panny.

mFT is an annoying premium market: you pay more for less: slower lenses by in large, and more per f stop under f3 than FT or sony for example. They use less materials, which is probably one reason they went FT rather than APSC or of course FF . Oly would have been well served with a manual FF based on OM lenses sans autofocus as a serious pro, top am' cam'. But they struggled to make margin out the top line from the OM2 apparently and made cheaper lenses after y2.

The japanese engineers have been at it since the 1950s: compete on size-miniaturise and use mechatronics to employ less labour per part and less materials too.

For me the cash on the lenses which flatter the body is just too difficult. I mean if they had a killer 12-40 f2-2.8 and a 40-150 f3.5-4 I would be happy to buy a complete system. Then I'd have the OPTION of the 45mm f1.8, not feel the necessity for faster glass.