Downsampling and Image Sharpening --Continuing from Lynne's thread

[ggt]

Hey all!

I got interested in delving more in to sharpening techniques with PSE after mangling Lynne's photo in the last "lens blur" thread she posted. I went thru some articles and books I had and the CandC Tutorial.

I just go the book recommended at the end of the Tutorial: Real World Image Sharpening. So far, so good. I think I can grasp most of it.

What confuses me is downsampling. I'll never understand the math involved, but HOW exactly does one downsample. Is it just reducing the dpi in the "resize" option on PSE. Like I have to for web publishing? I usually have my images at 300 dpi to play with and then downsize to 150dpi and change to .JPEG for ease of uploading.

I get the idea that mathmatically it is the opposite of compression. A % of each pixel is taken and reconstructed into the image. Is that right?

Also, what is the use of a .TIFF file. I think It's for archival purposes? Is that right?

THANKS SO MUCH!!!!

[Geoff F]

Trying to put things simply, Gretchen.

Reducing an image size is just entering new numbers in the size boxes, pixels or actual dimensions, but be careful that the resolution doesn't also change, or is at a suitable size for the intended purpose.

Resolution doesn't matter in many cases; it is just the total number of pixels which gets changed. Resolution is always variable.

However, if a fixed resolution is required, as with some printing companies, I usually set that first then adjust the dimensions. 300 ppi is often requested but with my own printer I can often get perfectly acceptable results as low as 200 ppi.

Printer dots per inch is something else which you don't need to worry about now.

When downsizing, it is sometimes recommended to use the Bicubic Sharpener setting; but I often find that I'm not happy with the results.

So after any substantial downsize I apply a little Unsharp Mask to suit each individual image. But an average for me would be 60 to 80% and around 0.5 pixels with 0 or 1 threshold level. Other people may use slightly different settings.

For web use, a resolution of 72 ppi is normally used with US monitors, but 96 ppi in Europe. But in reality this isn't important as everything will be auto set on a computer screen. But, having a fixed resolution, as required, will ensure that your image size ends up as you intended.

If you aren't careful you can get auto resized to the 'wrong dimensions' or undergo an auto resize with some software which loses quality.

So for normal internet use, I use 96 ppi resolution and a dimension of 700 to 1,000 pixels on the longest edge.

At one time it was recommend to resize in fixed percentage changes which only used specific numbers. But modern software can adequately cope with most sizes.

Tiff format is an excellent storage format which doesn't lose quality, like Jpeg, it can also retain layers etc when used with suitable software. I rarely use Jpeg except for internet use.

The downside of Tiff is very large file sizes. Therefore, I now save in the Photoshop psd format which produces smaller files but doesn't lose any quality.

The potential problem with psd is when sharing files with someone who has different software which doesn't recognise psd. I get that when using photos with my desktop publishing software.

In that case I have to resave as Tiff; but this usually isn't a problem because I also have to resize for my intended purpose.

[Glenn NK]

The mathematics of downsampling are beyond me, but this is an interesting article:

http://photographylife.com/why-downs...-reduces-noise

Or maybe it will make things more confusing.

Or this could help:

http://www.controlledvocabulary.com/...nsampling.html

Glenn

[Geoff F]

ps. If you still have problems understand pixels and resolution this is an odd way which originally enabled me to understand the principles.

Instead of pixels think about pixies. They each have a Red, Green or Blue hat. Adding more pixies increases the overall dimensions of the group. Removing some of them decreases it.

But the key is how far they stand apart (resolution). If too far apart gaps appear between them and their coloured hats produce a poor blocky looking effect when viewed from above.

You can even get them to change the colour of their hats to produce different pictures.

Silly I know, but this idea worked for me.

[davidedric]

Hi Geoff

I think I can hack the maths if I have to, but I love images. Thanks. Dave

[dje]

I get the idea that mathmatically it is the opposite of compression. A % of each pixel is taken and reconstructed into the image. Is that right?

Not really Gretchen. You cant take a % of a pixel, it's all or nothing .There are plenty of tutorials around on the web that describe down-sampling but basically you have to repace the set of pixels in the image with a new (smaller) set in which the new pixel colour values are estimated by looking at several of the original pixels in the same vicinity. Compression is an entirely different process.

As far as re-sizing in PS, if you are just doing it for display rather than printing, you are probably better off forgetting about dpi and just working on the Image Size in pixels. eg if your original image size was 4000x3000 pixels (a 12MP image), you could resize it to 1000x750 pixels and then save it as a jpeg.

Dave

[Dave Humphries]

Hi Gretchen,

Shoot with Large/Fine jpg, or better still RAW, so you get maximum resolution from the camera.
Your picture comes out of the size it is, I think you said '4254 x 2832' in the other thread, which sounds about right for a 12MP camera.

As is 4000 x 3000, Dave's shoots at. These numbers vary if you select different aspect ratios; 4000 x 3000 is plainly 4:3, where as yours is the standard 3:2, or if that's what any specific cameras is, but I digress.

Here's my take on it, not everyone will agree

If you're not concerned with producing an image file of a certain ppi for a stock company, do yourself an enormous favour and forget about dpi/ppi, and forget about physical image sizes in cm or inches too.

So you open it in PSE (or whatever), work on it and save an unsharpened version (different filename) without touching any re-sizing or dpi options - it will save at maximum resolution.

If you want a print file; first sharpen it, then save with different filename.

Now you want one for web use; assuming it is still open in PSE, undo that last sharpen, downsize it to 1600px width, or no more than 1000px tall, output sharpen it (my preferred values are 0.3px, somewhere between 80% and 110% and with 0 or 1 threshold), save this with a different filename too, perhaps a suffix indicating the size and orientation (e.g. _W1600 to indicate a 1600 wide landscape shot, or _H1000 would be a 1000px height portrait shot, or _S1000 for a 1000px square shot).

Always change the units (when re-sizing) to pixels, that way you won't get distracted by pesky (and unnecessary) ppi/dpi figures and sizes in cm or inches that mean little or nothing of relevance when viewed on screens of different sizes.

For web use, you generally know what size you are aiming for in pixels anyway, so why bother with cm, in, or even percent?

Cheers,

[rtbaum]

Trying to put things simply, Gretchen.

Reducing an image size is just entering new numbers in the size boxes, pixels or actual dimensions, but be careful that the resolution doesn't also change, or is at a suitable size for the intended purpose.

Resolution doesn't matter in many cases; it is just the total number of pixels which gets changed. Resolution is always variable.

However, if a fixed resolution is required, as with some printing companies, I usually set that first then adjust the dimensions. 300 ppi is often requested but with my own printer I can often get perfectly acceptable results as low as 200 ppi.

Printer dots per inch is something else which you don't need to worry about now.

When downsizing, it is sometimes recommended to use the Bicubic Sharpener setting; but I often find that I'm not happy with the results.

So after any substantial downsize I apply a little Unsharp Mask to suit each individual image. But an average for me would be 60 to 80% and around 0.5 pixels with 0 or 1 threshold level. Other people may use slightly different settings.

For web use, a resolution of 72 ppi is normally used with US monitors, but 96 ppi in Europe. But in reality this isn't important as everything will be auto set on a computer screen. But, having a fixed resolution, as required, will ensure that your image size ends up as you intended.

If you aren't careful you can get auto resized to the 'wrong dimensions' or undergo an auto resize with some software which loses quality.

So for normal internet use, I use 96 ppi resolution and a dimension of 700 to 1,000 pixels on the longest edge.

At one time it was recommend to resize in fixed percentage changes which only used specific numbers. But modern software can adequately cope with most sizes.

Tiff format is an excellent storage format which doesn't lose quality, like Jpeg, it can also retain layers etc when used with suitable software. I rarely use Jpeg except for internet use.

The downside of Tiff is very large file sizes. Therefore, I now save in the Photoshop psd format which produces smaller files but doesn't lose any quality.

The potential problem with psd is when sharing files with someone who has different software which doesn't recognise psd. I get that when using photos with my desktop publishing software.

In that case I have to resave as Tiff; but this usually isn't a problem because I also have to resize for my intended purpose.

Is there an advantage storing in Tiff or PSD vs. DNG?

[Colin Southern]

Is there an advantage storing in Tiff or PSD vs. DNG?

DNG is a RAW file format, so it's completely different to TIFF / PSD.

Just to expand on that a little ...

Photo editors like Lightroom and ACR are what's called PARAMETRIC editors in that they don't adjust actual pixels ... they only describe how the image should be presented by giving a long list of "instructions". These instructions can be included ("saved to") a DNG file, but the original data is never touched. Editors like Photoshop and PIXEL editors - they actually change pixel data - and the resulting file then needs to be saved in a format like PSD / TIFF / JPEG.

[ggt]

Ok, I think I"m understanding. I don't have much problem with the pixel concept or how the camera (in theory) captures and manipulates the photons to make the image. I can grasp the edges of that enough for my purposes.

If downsampling and downsizing are the same thing --yes, I think I can handle that too. The computer takes the average of the pixels in a certain radius and makes them into one pixel? (I know, such technical language -right?).

I always make a copy and work from the copy of the original. The originals stay in the iPhoto library and my work stays in Photoshop Organizer. Camera shoots in RAW, downloads to computer in ,JPG and I save in PNG. TIFF is confusing me because I"ve never been offered the option with my programs. --but I've never asked it to either.

If I am done with an image and want to save it to my archive drive --should i do it in TIFF?

[John Morton]

I always make a copy and work from the copy of the original. The originals stay in the iPhoto library and my work stays in Photoshop Organizer. Camera shoots in RAW, downloads to computer in ,JPG and I save in PNG. TIFF is confusing me because I"ve never been offered the option with my programs. --but I've never asked it to either.

If I am done with an image and want to save it to my archive drive --should i do it in TIFF?

Hi, Gretchen;

I'm a long time and die hard user of TIFFs, so I'll tell you how that came to be and perhaps you'll get an idea of how that file format can works for people, from the context of my experience.

I started into a continuous use of digital images when I bought a dedicated 35mm film scanner - a Minolta DiMAGE 5400 Scan Elite. This was just when it first came out so that would have been about 2004. Since I was interested in scanning a large number of my 35mm film archive, and didn't want to be rescanning images when I could be scanning new material, I scanned everything at the highest settings available. This meant doing 16 bit scans; and that meant saving the files as TIFFs.

Then, I could take the original, high quality scan and do anything I wanted to them (downsizing them as needed, for instance).

(I did use a little Logitech handscanner on hard copy photographs back in the early 1990's, and a Microtech flatbed scanner by about 1993, but I wouldn't really count that nowadays).

Working with 16 bit images also limited the number of photo editors I could work with: Paint Shop Pro, or, Photoshop were pretty much my only options. I tried both, and went with Photoshop 7 (the version prior to the first Creative Suite release).

Four years later, switch to digital capture from film (a Nikon D700), I started using Nikon's RAW files (NEF) as my primary image format. But, I still use TIFFs for all my image editing; and here are a few of my reasons:

1) Compatability: TIFFs are a well established and universally accepted file format. I can output a TIFF from a NEF RAW file using Nikon's Capture NX2 software and I can work on that file(s) in a number of programs (such as HDR creation software) and when I go to Photoshop, the TIFFs I have of any image will still align perfectly should I want to combine an HDR image with one of the original TIFFs it was produced from. I can't be sure of that if I am using RAW files because they might be rendered differently in different software; and I can't be sure of that using JPEGs because I can't be sure the compression algorithms of different software will match (others may correct me on what is perhaps an anti-JPEG superstition).

2) TIFFs are bitmapped image files, which means that each pixel has an exact location on an X/Y axis (like squares on a sheet of graph paper). That makes for larger image files, yes, but it also ensures optimum image quality because the pixel locations and values are stable and certain. In JPEGS, the image is defined by equations that group pixels together in (at least) blocks of eight; so the compression of the JPEG means that the composition of the image is actually somewhat fluid at the pixel level.

3) All of the software that I have will work on 16 bit TIFFS; but then of course, that's probably also because I won't purchase software that won't work with those files. TIFFs are a file format that I can bring into any software; as opposed to Photoshop's PSD format which doesn't give me that option. Nikon's NEFs are also widely accepted, since Nikon has made their specs openly available to software developers; but again, I don't want to use a format that is subject to conversion as a basis for editing images in different programs when I might want to merge the different results later and I will want the different results to still align and register properly with each other.

4) So as a result, my edited images end up being saved as TIFFs; but as far as "archiving" goes, I consider my original NEF files in RAW format to be my ultimate archive. But, I only use JPEGs for posting on the Internet and for emailing; so when I produce a JPEG from a TIFF, iti s generally a one-off image file used for a specific end.

I'm not sure what you mean when you say that your camera shoots in RAW but "downloads to (your) computer in JPG"... do you mean that all of your files start out in your computer as JPEGs? If I have a JPEG from somewhere, anywhere, that I am going to work on in editing then the first thing I do when I get it onto my computer is to re-save it as a TIFF...

[dje]

I always make a copy and work from the copy of the original. The originals stay in the iPhoto library and my work stays in Photoshop Organizer. Camera shoots in RAW, downloads to computer in ,JPG and I save in PNG. TIFF is confusing me because I"ve never been offered the option with my programs. --but I've never asked it to either.

If I am done with an image and want to save it to my archive drive --should i do it in TIFF?

Gretchen I think, as John has suggested, we need to clarify what you mean by "camera shoots in RAW, downloads to computer in jpeg".

I suppose you could say that the camera always "shoots in RAW" in the sense that the output from the camera sensor is the RAW data. But the terminolgy is usually used as follows

• Shoot in RAW : camera is set to save image in RAW format and this RAW file is downloaded to your computer.
• Shoot in RAW + JPEG : camera is set accordingly and saves the image as both a RAW file and a JPEG. Both can be downloaded to you pc.
• Shoot in JPEG : Camera is set to JPEG and image is saved as a jpeg and downloaded as a jpeg.

Which mode do you use ? I ask because it has a bearing on how you save your files.

Dave

[revi]

(...)
I always make a copy and work from the copy of the original. The originals stay in the iPhoto library and my work stays in Photoshop Organizer. Camera shoots in RAW, downloads to computer in ,JPG and I save in PNG. TIFF is confusing me because I"ve never been offered the option with my programs. --but I've never asked it to either.

If I am done with an image and want to save it to my archive drive --should i do it in TIFF?

I see no need to use TIFF for that, PNG works just fine to store a final image in. I use 16-bit PNG all the time for final images. For intermediate saves, it's either PNG, or the native format for my editor if I need to preserve layer information and such (which PNG can't handle, but see below)

To take John's points (in shortened version):
PNG is a standard format, with less variations than TIFF, and can store 16 bits/channel, including transparency (not very useful in photo's, I admit)
PNG is also a bitmapped format, so no problems with alignment due to the format. It also stores colour space information, so rendering should not pose any problems either.

Both TIFF and PNG allow lossless compression, but I still find that TIFFs take up twice as much place as PNGs.

The one thing that PNG cannot do, is store image layers. But, you don't need that for finished images, and if I need intermediate storage of a work in progress, I use the native format for my editor (.xcf for the GIMP, for photoshop that would be PSD), to keep a maximum of information for the next editing session. Main disadvantage: those files can get rather large. (Note that although TIFF can store layers, I'm not sure it would store all the information there is in a PSD or .xcf file, which is why I avoid using TIFF for that case; the TIFFs would also be large anyway).

I can see why John started working with TIFF (it was about the only valid option available when he started using them), but I cannot see any reason to switch from PNG to TIFF (nor the other way around). Both are standard formats that do what they need to do.

@John: you cannot compare TIFF and RAW for several reasons. RAW isn't one format, but a series of formats. More important, it's not conceived as an image format. For one thing, you absolutely need the metadata to be able to interpret it correctly: matrix layout, white balance, and such are essential. Nor is it supposed to be rewriteable.
With respect to the register problems you mention in relation to HDR, those shouldn't occur with either TIFF, PNG or RAW.

[Alison Johnston]

Worth a watch

http://youtu.be/44mV3NsLmXw

[ggt]

Dave, as best as I understand it the EXIF shows the picture was captured in RAW format. When It shows up in iPhoto it is a .jpg. I am understanding i would be happier if it was a .TIFF. Can I change this>

[Andrew76]

Gretchen, taken directly from apple.com:

When you import a RAW image to your iPad, iPhoto will display only the JPEG version of the image embedded in the RAW file. When editing a RAW image in iPhoto, the edits are derived from the embedded JPEG, and saved in JPEG format.

It seems as though iPhoto cannot handle RAW photos in the same way as a dedicated RAW processor like ACR or LR.

But I am not a Mac user, and cannot help you any further - I just know that ACR, and LR have always worked for me, and if I am shooting RAW, I can edit RAW, and save them however I like.

[dje]

Gretchen I'm not an Apple user so I'm not familiar with iPhoto. But given what Andrew has said, things are starting to make sense. You can tell what capture mode your camera is set to by looking in it's menu system. It will be RAW, RAW+JPEG or JPEG.

I haven't used Photoshop elements for some time now but I think you can open raw files from PSE Organiser into Adobe Camera Raw. This is probably the way to go for you if you want to process the raw file. But you may have to look at how you import your images from the camera to make sure the raw file is gettting into Organiser. Calling on Apple experts for help here please !!

If you are working with a jpeg as your "original" or "master" file, I personally cant see any point in converting it to a TIFF or PNG at the initial stage. This wont improve the quality of it. What you need to avoid however when working with jpegs are multiple Save As jpeg's in series. Each time you re-save as a jpeg there will be a degradation in quality, the size of which will be determined by the quality settings used.

So when you have done some editing in PSE, save it either as PSD, PNG or TIFF so that you can come back later and modify the edit without losing any quality due to another jpeg compression. Save a separate jpeg of your work for publishing on the web when required.

Dave

[Dave Humphries]

Worth a watch

http://youtu.be/44mV3NsLmXw

Thanks Alison,

I particularly enjoyed the bit where he recommends slapping anyone saying you need 72 dpi for web use

[dje]

Thanks Alison,

I particularly enjoyed the bit where he recommends slapping anyone saying you need 72 dpi for web use

Yes I thought it was quite a good video too. I must have a look at some of his other videos.

Dave

[ggt]

Andrew, thanks for looking that up. I"m a bit upset now with iPhoto. As all my RAW files imported from the camera and then deleted have been convered to .jpg. I am assuming (I know, never a good idea) that they have been compromised as a result--even tho they get again converted to .png when I import them into photoshop.

Is that right?

I have to investigate bypassing iPhoto and just opening them in PSE. First I have to make sure it will open in RAW--if I am reading Dave Ellis correctly.

AAAAHHH, I hate being a newbie.

When I save a photo for web publishing, I have a separate file prefix I use so I know that is why I saved it. Usually web20120000(the date)ggtnameoffile.jpg. Everything else is dateinitialsfilename.psd. It's a PITA to locate files, so I devised the code for searching. Now I understand why --iPhoto will not open a .psd. Since I only use it for importing and initial viewing (to decide whether to purge or import into PSE. It also archives old photos and those family members send me.

I know, my organization system is a mess. But it works for me.