moonlight HDR - Creative Uses of CHDK - CHDK Forum supplierdeeply

moonlight HDR

  • 14 Replies
  • 18008 Views
*

Offline fbonomi

  • ****
  • 469
  • A570IS SD1100/Ixus80
    • Francesco Bonomi
moonlight HDR
« on: 13 / June / 2008, 19:24:19 »
Advertisements
My attempt at a moonlight HDR...

The front trees are illuminated by lights on my house, the further landscape is purely moonlight

I have taken 10 65" frames in RAW format . Unfortunately, I deleted the dark frame so I couldn't subtract it :-(

These 10 frames are combined with imagestacker to create 4 different shots

1) average of 10 frames (i1+i2+i3 ... + i10)/ 10
2) sum of 10 frames (i1+i2+i3 ... + i10)
3) and 4) intermediate steps (i1+i2+i3 ... + i10)/ 4 and (i1+i2+i3 ... + i10)/ 2

then tonemapped in photomatix.

« Last Edit: 14 / June / 2008, 03:27:57 by fbonomi »

Re: moonlight HDR
« Reply #1 on: 13 / June / 2008, 22:33:56 »
Now that you have the luminance, try adding an overall blue cast to make it feel more like a nighttime shot.

The proverbial "day-for-night" has been turned inside out and you've made a "night-for-day!!" 8)
~~~~~~~~~~~~~~~~~~~~~

<")%%%><<

*

Offline Coutts

  • *****
  • 538
  • www.flickr.com/couttsphotog
    • Flickr
Re: moonlight HDR
« Reply #2 on: 13 / June / 2008, 23:04:48 »
dang nice, im still a bit confused about this whole dark frame subtraction thing and raw sum and average and im running in circles now lol.
Canon 5d
Canon 50mm f/1.8
Sigma 24mm f/1.8

Flickr

*

Offline fbonomi

  • ****
  • 469
  • A570IS SD1100/Ixus80
    • Francesco Bonomi
Re: moonlight HDR
« Reply #3 on: 14 / June / 2008, 02:54:08 »
Quote
Now that you have the luminance, try adding an overall blue cast to make it feel more like a nighttime shot.
The problem is, it all gets VERy un-natural, as even if  Imake blue-ish it looks like a blue-ish day, ot like a normal night ...

Quote
dang nice, im still a bit confused about this whole dark frame subtraction thing and raw sum and average and im running in circles now lol.


ok, quick explanation of stacking and averaging...

Suppose you have a black and white camera, 1 pixel x 3 pixels

An image would be made of three pixels, each pixel would have its luminosity expressed as a number (from 0 to 255).

A completely black image is something like this:
0 0 0

A completely white image is something like this
255 255 255

A well exposed image has a good range of values, eg
32 95 145

A night-time shot is very dark:
2 4 8

If you expose more (let's say 10 times longer), you get more light.
20 40 80

The problem is, the longer you expose, more "noise" you have (the numbers have small errors in them)
23 39 75

But, if noise is really noise, it's random, (sometimes giving a higher value and sometimes a lower value). S0, let's take 4 identical shots of the same image
a 23 39 75
b 19 35 79
c 22 44 83
d 18 42 80

If we do an average of each pixel, we can hope to avoid some noise and have a better image:
avg: 21 40 79 (averages of each pixel)
That is closer to out "theoretical" 20 40 80 image, and about of the same luminosity

If, on the contrary, we just add each column, errors get again summed up (so, hopefully they compensate for each other) but the image is obvioulsy much brighter: this is what is called stacking
sum: 83 160 317

In a real case, 317 is over out range of 0-255, so it stays to 255.
sum: 83 160 255
(this image is then partailly over-exposed)

In real cases, you have more than 3 pixels (you have a few millions of them) and you have more than a number per each pixel (actually, you have 1 for each RGB channel) but that's the idea.










« Last Edit: 14 / June / 2008, 15:13:20 by fbonomi »


*

Offline fbonomi

  • ****
  • 469
  • A570IS SD1100/Ixus80
    • Francesco Bonomi
Re: moonlight HDR
« Reply #4 on: 14 / June / 2008, 03:18:39 »
Practical example of averaging.

This is one of the frames I took:



It's a bit dark, so I will do "Levels" on it:



It looks better, but let's look at a detail:



A lot of noise.

Ok, then I take 10 different images and do the average. The image looks very similar:



I do levels on this image too, and already looks much better:



If you look look at a detail:



You see a huge difference: the noise was very strong in the original image, and here it hase been leveled out.

In the original image, the noise was almost as strong at the signal, so I couldn't lighten the image or I would lighten the noise too. Here i could (if I wanted) lighten the image much more and still have a lower noise level



« Last Edit: 14 / June / 2008, 03:21:18 by fbonomi »

*

Offline Coutts

  • *****
  • 538
  • www.flickr.com/couttsphotog
    • Flickr
Re: moonlight HDR
« Reply #5 on: 14 / June / 2008, 07:06:04 »
ah thank you very much :-)
Canon 5d
Canon 50mm f/1.8
Sigma 24mm f/1.8

Flickr

*

Offline fe50

  • ******
  • 3147
  • IXUS50 & 860, SX10 Star WARs-Star RAWs
    • fe50
Re: moonlight HDR
« Reply #6 on: 14 / June / 2008, 09:01:21 »
Hi fbonomi,
you gave me (us) a great explanation, thx.

What was the ISO for the pics ?
We have full moon here next wednesday, so i'll take my canon and my mini tripod with me to my night walk...

*

Offline fbonomi

  • ****
  • 469
  • A570IS SD1100/Ixus80
    • Francesco Bonomi
Re: moonlight HDR
« Reply #7 on: 14 / June / 2008, 09:55:43 »
What was the ISO for the pics ?

ISO 80, I didn't want high sensitivity to add to the noise..


Re: moonlight HDR
« Reply #8 on: 14 / June / 2008, 12:10:54 »
Excellent explanation! 8)
~~~~~~~~~~~~~~~~~~~~~

<")%%%><<

*

Offline Joshatdot

  • *
  • 20
  • A720 & A570 w/latest Autobuild
    • Joshatdot's Photobucket
Re: moonlight HDR
« Reply #9 on: 17 / June / 2008, 02:01:40 »
How do you do this...is there a FAQ/Wiki about this stuff?  I tried DeepSkyStacker, but I don't get clean, smooth looking pics as yours.



I am still trying to figure out DSS, I guess I need some darks, flats, dark flats, offset/bias images...

I have 10 images made with Non-stop Long Exp Timelapse and that's it.

Here's the same stuff done in PS CS2, all images were 'Screen Blending'

« Last Edit: 17 / June / 2008, 02:08:24 by Joshatdot »

 

Related Topics