Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Jonathan Wienke on April 06, 2008, 02:34:36 pm
-
Original Image:
(http://visual-vacations.com/images/2007/2007-12-24_0009a.jpg)
Upsized 200%, Bicubic Smoother + Add Noise, 20% Gaussian Monochromatic:
(http://visual-vacations.com/images/2007/2007-12-24_0009b.jpg)
Gaussian Blur, 1 Pixel:
(http://visual-vacations.com/images/2007/2007-12-24_0009c.jpg)
Add Noise, 2.5% Gaussian Monochromatic:
(http://visual-vacations.com/images/2007/2007-12-24_0009d.jpg)
By adjusting the parameters of the noise and blur, you can alter the characteristics of the grain pattern however you like. This technique will allow you to print as large as you like without obvious pixelation or other digital artifacts.
-
Cute trick, Jonathan.
Then how about spilling a little fixer on parts of the print to add the familiar incompletely-washed aroma and eventual yellowing that go with traditional materials.
-
By adjusting the parameters of the noise and blur, you can alter the characteristics of the grain pattern however you like. This technique will allow you to print as large as you like without obvious pixelation or other digital artifacts.
Glad to hear you are safe and still helping people out in photography!
Using each of your 3 examples as layers and adjusting the blending of each works to arrive at the look one wants easily.
Asher Kelman
-
Glad to hear you are safe and still helping people out in photography!
Using each of your 3 examples as layers and adjusting the blending of each works to arrive at the look one wants easily.
Asher Kelman
[a href=\"index.php?act=findpost&pid=187465\"][{POST_SNAPBACK}][/a]
When I need to add grain, I tend to add it on a layer using the DXo filmpack plugin for CS3. Usually I use the tri-x pattern and adjust the size and intensity to suit. Also, I hope to avoid it being obvious in the final print - unless I want it to be for old times sake...
Mike
-
When I need to add grain, I tend to add it on a layer using the DXo filmpack plugin for CS3.
The point of this exercise is to explore what's possible with the tools already in Photoshop.
-
You can take things even farther:
First, one can easily apply your approach to a 50% gray layer in overlay blend mode. This however gives you very little control over the amount of grain added to the various luminosities. You won't get any grain in the highlights for example...
Take DXO's Filmpack: It uses, as described on their website, somehow an intensity function which depends on the luminosity.
To simulate this, I use two curves adjustment layers. One for the light (blue curve), the other for the dark (black curve) grain.
In this example the numbers for the (x,y)-points of the curve are:
Light grain: (0,16), (100,150), (255,240)
Dark grain: (16,0), (150,100), (240,255)
[attachment=5972:attachment]
Clearly these two curves annihilate each other (which is mandatory) without layer masks.
One needs some appropriate layer masks. Appropriate means here that the mean luminosity value of the layer mask must be around 127. You can take scanned grain, artificially created etc.
Now just put the grain mask to one and the inverted grain mask to the other curve.
So now you have total control over the responding behavior of your artificially created film; you can add further control points, put the two curves in a layer set and control the strength etc.
All non destructive.
There are several points where this technique can be pushed farther. Add color noise, refine the grain creation with some more complex approaches etc...
-
The point of this exercise is to explore what's possible with the tools already in Photoshop.
[a href=\"index.php?act=findpost&pid=187496\"][{POST_SNAPBACK}][/a]
Sorry, didn't realise. Just that I prefer the DXo 'grain' to photoshop noise - like the blur approach though and the curves suggestion.
Mike
-
For those of you whe care, I created an action that automates this process and saves the results in layers so you can adjust the intensity of the effect by tweaking the layer opacity. The default settings are designed to disguise upsizing artifacts as well as possible without affecting image detail any more than absolutely necessary. Here's before and after samples:
(http://visual-vacations.com/images/2007/2007-12-24_0009a.jpg)
(http://visual-vacations.com/images/2007/2007-12-24_0009e.jpg)
You can download the action here (http://visual-vacations.com/Photography/MiscFiles/FilmGrain.zip).
-
For those of you whe care, I created an action that automates this process and saves the results in layers so you can adjust the intensity of the effect by tweaking the layer opacity. The default settings are designed to disguise upsizing artifacts as well as possible without affecting image detail any more than absolutely necessary. Here's before and after samples:
You can download the action here (http://visual-vacations.com/Photography/MiscFiles/FilmGrain.zip).
[a href=\"index.php?act=findpost&pid=187678\"][{POST_SNAPBACK}][/a]
Thanks for the action.
It looks similar to the technique presented by Michael in the "From Camera to Print" tutorial using Pixel Genius tools.
Cheers
Julien
-
I made a small change to the action; it upsizes using Bicubic Sharper instead of Bicubic Smoother now. This slightly increases the sharpness of the output image without increasing artifacts. The sample image and the action file download have both been updated.
-
It is clear that your before image is sharpened for display output. You get some pretty strong halos around the edges of the tree after upscaling.
Can you explain to me, why you are blurring the whole image after adding the noise? Especially because you seem to be so concerned about sharpness...
Also I don't get why you try to reinvent the wheel. You might consider reading Schewe's article upres article (http://www.digitalphotopro.com/tech/the-art-of-the-up-res.html).
Although the noise section is pretty basic, the sharpen and upscaling section really gets down to the point. Schewe rightfully uses bicubic smoother. After that, he applies Photokit's Super Sharpener, which is the same method described Fraser's sharpening book as Texture Brush. In principle it is just a sequence of USMs with decreasing radius and increasing increasing amount.
Result:
[attachment=5996:attachment]
-
Can you explain to me, why you are blurring the whole image after adding the noise? Especially because you seem to be so concerned about sharpness...
I'm not. I'm only blurring one of the noise layers to filter Photoshop's noise function so it takes on the appearance of larger, organic clumps instead of the exactly-pixel-sized noise particles you usually see from Photoshop.
Also I don't get why you try to reinvent the wheel. You might consider reading Schewe's article upres article (http://www.digitalphotopro.com/tech/the-art-of-the-up-res.html).
Although the noise section is pretty basic, the sharpen and upscaling section really gets down to the point. Schewe rightfully uses bicubic smoother.
I use Bicubic Sharper because it brings out more image detail without increasing the noise level, unlike your example. I tried both, did the comparisons, and chose Sharper based on the results. I know it's technically "backwards", but I'll take results over theory any time.
One of the design principles of the action is to leave the after-action image in a state visually equivalent to the original--if you compare the after-action image at 50% to the before-action image at 100% in Photoshop, they will be visually indistinguishable. This means that noise levels need to be kept low (to keep detail from being obscured), and tonality and overall sharpness are not altered.
Your version of the image is sharper than mine, but you've also sharpened the noise well beyond what is needed to disguise upsizing artifacts. If you want to do additional sharpening after upsizing, do it only to the the original image layer under the noise layers. That will accentuate true image detail without simultaneously sharpening noise.
After that, he applies Photokit's Super Sharpener, which is the same method described Fraser's sharpening book as Texture Brush. In principle it is just a sequence of USMs with decreasing radius and increasing increasing amount.
I've been using that as a general sharpening technique since 2003, see http://www.outbackphoto.com/workflow/wf_20/essay.html (http://www.outbackphoto.com/workflow/wf_20/essay.html) for a description of the my original action set, and http://www.visual-vacations.com/Photograph...ningActions.htm (http://www.visual-vacations.com/Photography/SharpeningActions.htm) for a more recent version. Lately I've been using Focus Magic for deconvolution-based capture sharpening, and then my midtone actions for creative sharpening.