22MP H3D photo enlargement question...

[frjonah]

I am wondering what experience forum users have enlarging 22MP files from Hasselblad H3D cameras. The default print size (at 300 dpi) is around 13 x 19", but I know I can get more out of the files than that just by looking at the level of detail in Photoshop.

We have a Canon ipf8300 (an absolutely amazing wide format giclee printer!) that prints up to 44" wide that we use for our core business that I'm going to do some sample prints on, but since such prints are a bit costly to produce, I'd like to make sure I don't go too far with my upres-ing for the initial trials.

What is a good size you would expect to be able to enlarge to with a 22MP shot (using PerfectResize - formerly Genuine Fractals)? How about your maximum... how far have you pushed 22MP shots with acceptable results?

I have a follow-up question as well: can you get bigger enlargements from film than from a 22MP back? I have the HM 16-32 film back and some velvia slide film and I'm looking forward to doing some experimentation, but some initial direction would be most helpful.

Many thanks!

Fr Jonah

 

[DPaul]

Hi,

I have done the following test:

I printed the same pic on an epson 3800 at 300,250,200,150,100 dpi.

Then you scramble the prints and guess which is which.
Very, very difficult to see a difference. I would say you can co easily to 200 dpi size.

Paul

 

[frjonah]

Thanks for the response, Paul... that sounds like a test worth doing. What size prints were you doing in your test?

 

[DPaul]

Thanks for the response, Paul... that sounds like a test worth doing. What size prints were you doing in your test?

Hi,

There is no need for large prints , I used half-page (A5) sheets.

I used 2 different techniques, starting from a pic that is tack sharp with lots of details. You need PS or PSelements.

1)
a) Cut a 300 dpi A5 rectangle from that pic. (about 2520 x 1770 pixels)->print
b)Without resampling in "image size" you modify the 300 into 250.
Your pic would now print larger, but we use the crop tool to cut out the middle piece of the pic with the same A5 dimensions as (a). Easy!->print
c) Do (b) for 200,150,100.
Your pic will be A5 size all the time, but show the centre of the image in ever greater enlargement. That is the same effect of printing bigger!
You'll be amazed. (I suggest a blind test with your "expert friends", fun guaranteed)
2)
a) Start as 1(a)
b) But now reduce the dpi while keeping the dimensions (A5) unchanged. (With resampling)
So changing the dpi will reduce the pixels automatically (300,250,200,150,100)
So identical image, same size, different DPI;

Make sure to write on the back what the dpi is, or you may never get them in the right order again!

Paul

 

[frjonah]

Paul,

Your ideas are very intriguing. I also professionally print our Orthodox Icon images using a Canon ipf8300 (max print width = 44"), which uses pigment based inks... it's really a dream to use and the prints, well, I couldn't be happier on so many levels. Our actual products consist of canvas prints mounted to traditional solid wood panels with slats (such as those used in antique Russian icons, if you're familiar), so I think the test(s) will likely be the most representative if we print on canvas. That said, you can get by with a bit more on canvas than you can on photopaper, so I think I'll conduct the test on photopaper and use the resulting enlargement thresholds I arrive at for canvas as well, knowing that it will skew things a bit in the direction of higher quality (always a good thing!).

Your trial descriptions also gave me another idea, which I'm thinking will directly show how large an enlargement percentage to standardize to on a typical image using Perfect Resize (formerly Genuine Fractals). This will be upres'ing 22MP shots from my Hassy H3D-22.

What do you think of this approach:

1. Take a full size image of one of the images I recently captured (set the image dpi to 300 with no resampling) and processed for reproduction and cut out a 3 x 5" section in the center of the face of one of the figures.
2. Upres the image 20% using Perfect Resize and cut out another 3 x 5" of the new print.
3. Upres another 20% and repeat.
4. Continue this process until I can see visible pixelation on the screenshot at 70% zoom.
5. Layout all these 3 x 5" swatches on a 44" Photoshop canvas and print.
6. Critically examine the resulting series to determine the threshold for unacceptable pixelation.
7. Drop to 20% below that level and call it standard for perfectly exposed / perfectly in-focus shots.

What do you think about this process?

In any case, I'll try your two ideas and mine and report back on here upon completion with my findings.

This is exciting!

Fr Jonah

 

[DPaul]

Hi,

There are many ways to...

In the methd you propose, the DPI variable is kept constant at 300 I suppose.
This implies that your Upres procedure adds pixels to the pic that have been "invented" by the software. (Invented = intrapolated)

There are softwares that are very good at this, but still i am very curious to see the difference (if any) between prints of the same size:

a) Upres 30% at 300 dpi, thus adding pixels
b) changing the dpi from 300 to 200 = same size as (a)

Paul