Can we have a 15 x 2.5 metre image?

[PCPix]

Advertisement (gone after registration)

A bit of general interest for the SL2 shooters:

I recently received a call from a potential new client asking if I had any New Forest panoramic images, as they needed something for their stand at the Chelsea Flower Show.

As the conversation progressed, the client revealed that they needed a single image backdrop 'wallpaper' behind their topiary display of life-sized New Forest animals.

And it would need to be 15 x 2.5 metres in size.

And they would like heather in the foreground, trees in the distance and blue sky, preferably in 'thirds'.

And can they have it by next Tuesday (four days away)?

Nothing like a deadline and a strong 'wish list'...

It became clear that I would need to shoot a specific image to fit the format (extremely narrow) which would ideally be of a flat landscape - as undulations would upset the third/third/third proportions. Weather forecast wasn't great, but very early Sunday morning looked to have a couple of clear hours and there was one possible location which still had a good flourish of heather and an almost flat horizon.

The company printing the backdrop requested 100dpi as the viewing distance could be as little as 1.5m. A quick 'new document' in photoshop revealed a 1.6Gb filesize!

The only possible ways to get anywhere near this amount of photographic data with my current equipment was to use the SL2 with a 4-shot stitch from 'Multi-shot' originals in landscape format or a 12-shot stitch from 'regular' portrait format originals.

On the morning of the shoot, conditions were perfect, clear sky, practically no wind and a light dew accentuating the colour in the heather. Lens choice was 75mm APO Summicron M, set to f/11. The depth of field does deliberately fall off in the foreground, allowing the 'real' stand display to visually blend into the background (see images 5 and 6)

I did shoot both options for stitching, however the 12-shot 'portrait' stitch ended up giving the better result. All images were processed from the DNG files into 16-bit files and then assembled using Photoshop's photomerge (using blend and vignette removal options). The end result was a 3.4Gb 16-bit file which required very little additional work, a minor amount of upsizing (effectively to get the height of the image up to 2.5m at 100dpi). Finally downsampled to 8-bit and sent to client as a 1.6Gb Tiff file.

The client was awarded a "Silver Gilt" award for their stand - a truly wonderful achievement for a first show at Chelsea.

I can make no claim to their well-earned success, however I will lay claim to being 'in the background'... 😉

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[helged]

Very nice! In the last image, the deer is almost a part of the background 👍. 

[satijntje]

Wow, what a challenge!
You did very well!

Congratulations

[wjdrijfhout]

This is an amazing result. And how the foreground objects blend in with background is just perfect. So final size was around 600MP. Typical use-case of SL2 over SL… 

Just curious, in comparing 4-shot panoramic multi-shot vs 12-shot portrait, this was using different lenses, I presume?

[brembeck43]

Super!!!

[PCPix]

42 minutes ago, wjdrijfhout said:

Just curious, in comparing 4-shot panoramic multi-shot vs 12-shot portrait, this was using different lenses, I presume?

This is a great question - at first glimpse it would seem straightforward and easier to shoot the 4x multi-shot.

My lens choice was: 75mm APO Summicron-M, 50mm Summicron-M (v.3) or the Panasonic 24-105.

50 was perfect for the landscape 4-shot, 75 for the portrait 12-shot and the 24-105 could be zoomed to do either.

All would be shot at f/11 to try for the best resolution and depth of field compromise - corners of all (especially the 24-105) can be ignored as the overlap for stitching effectively excludes any deficiencies.

Knowing the final use, I did favour using one of the M-lens options.

I think the easiest way to explain it is that the 4x multi-shot file is made up from four image circles and the 12x regular is made up from twelve image circles. Each landscape original is roughly equivalent to 3x portrait originals. Each multi-shot generates more digital information, but the 3x portraits generate more optical resolution.

I did a test of both options in the field and the portrait orientated images looked better on the camera back - so the decision was made on the spot to do it that way.

Bearing in mind this was being shot to a specific brief and to a short deadline, my choices were based on ensuring the best result to fit the client's brief and usage. The ulltimate would have been to do 12x portrait multi-shots, however this may be a project for another day 😉

[hofo100]

Advertisement (gone after registration)

Very interesting, thank you. Just as a matter of interest - did you consider the PS/LR super resolution function to arrive at the necessary mp? And no issues with your processor getting the work done?

[PCPix]

38 minutes ago, hofo100 said:

Very interesting, thank you. Just as a matter of interest - did you consider the PS/LR super resolution function to arrive at the necessary mp? And no issues with your processor getting the work done?

I'm still on CS6/Mac - so generated the data in camera, simply to be stitched in PS. No problems with processing (about 3 minutes for the stitch with 16-bit files)