snubbr.com

Resampling quality
Hi all,.

If one compresses an image (JPEG) then pixels are 'merged' to reduce the file size. If one resamples an image (say 10 Mp) down to 640 by 480 for emailing then one assumes that there must be a loss of resolution also?.

I took one of my first digital photos at 5 Mp and then reduced the size for emailing. In the picture was a wooden chair which looked OK at 5 Mp but at circa 640 by 480 the straight edges (wood) looked like a 'saw blade' they were so badly resampled. Not sure if this was the software or if it's just and expected artifact of resampling to lower resolution..

So, my question is simple - is it better to set the camera at 640 by 480 if one intends to email the picture(s) or is it better to use software to reduce the picture size?.

I'm not sure if the camera uses internal software to reduce the file size or the sensor is just 'adjusted' to capture only the required (640 by 480) pixels?.

Regards,..

Comments (9)

Antony John wrote:.

So, my question is simple - is it better to set the camera at 640 by480 if one intends to email the picture(s) or is it better to usesoftware to reduce the picture size?.

Usually it is better to set the camera to maximum resolution and reduce the size afterwards using software..

It's likely that cameras from different manufacturers use slightly different internal methods, but in many cases the results at less than the full resolution can be quite ugly, with staircase artefacts in diagonal lines..

The software, for example Irfanview, may offer more than one method of reducing the size. Resampling is definitely recommended, as it reduces the harsh stepped edges you noticed.http://www.irfanview.com/.

Of course most image editors can do the resizing too. But be careful to select "resample" rather than "resize"..

I'm not sure if the camera uses internal software to reduce the filesize or the sensor is just 'adjusted' to capture only the required(640 by 480) pixels?.

I don't think the sensor can be configured differently. This is done by software inside the camera, and sometimes (often?) is not done very well..

Regards,Peter..

Comment #1

Antony John wrote:.

I took one of my first digital photos at 5 Mp and then reduced thesize for emailing. In the picture was a wooden chair which looked OKat 5 Mp but at circa 640 by 480 the straight edges (wood) looked likea 'saw blade' they were so badly resampled. Not sure if this was thesoftware or if it's just and expected artifact of resampling to lowerresolution..

You do not say HOW you reduced the image size, and you do not say HOW you judge the result. If you judge the result by looking at the image on screen, blown up to the same size as the original (for example filling your entire screen), you will obviously see those 'jaggies'. You have less pixels to make up the same image, so the pixels are shown bigger if you view the image at the same size. The same effect will show if you print that smaller image..

However, if you look at the resized image at 100% (meaning it won't fill your screen any longer), there is no reason to see 'saw blade' lines, unless you resized using extremely poor software. Look at the average photo in the so many web galleries you can find on the internet. They don't show saw blade lines..

Johanhttp://www.johanfoto.com..

Comment #2

He photo I referred too was taken 4 years back with a Sony 5 Mp DSC camera and processed with supplied Sony software..

What matters more is which is the better way to take photographs that will be viewed at 640 x 480 (or 600 x 800) as email attachments:1) Reduce the image size using the camera's resolution settingor.

2) Capture the picture at full resolution and resample down to the required file size..

I have seen on the internet (and have books that cover the subject) on how compression works but find no reference to how resampling works..

I find that resampling results in less sharpness even when viewed at the native resolution onscreen (i.e. actual pixel size). This is expected using compression because of the algoritms used..

BTW, I now use a D80 with PS Elements 4...

Comment #3

Antony John wrote:.

I find that resampling results in less sharpness even when viewed atthe native resolution onscreen (i.e. actual pixel size). This isexpected using compression because of the algoritms used..

BTW, I now use a D80 with PS Elements 4..

Just to reiterate what has already been said, the camera always takes the shot at full resolution(*); it is then resampled by the software (firmware) in the camera to the smaller size. In that sense it is exactly the same process as setting the camera to maximum resolution then doing the resampling on the computer. The difference is that Photoshop is much more powerful than the camera's firmware, and maybe even more importantly you have more control..

It is true that resampling makes images soft. This is inherent in the process - by changing a block of nine pixels (say) into one, you are averaging their values and therefore softening the detail. The resampling algorithms are much more sophisticated than a simple average, but the principle still applies. So along with resampling, you will always need to sharpen. (The camera does this too.).

I only have full Photoshop and I don't know where Elements differs, but I can choose from several resampling algorithms, one of which is Bicubic Sharper which is specifically targeted at downsampling. Personally I prefer to use the standard Bicubic method followed by separate sharpening because it offers more control..

Sometimes you may find that downsampled images benefit from slightly higher contrast and brighter midtones - adding 'punch' as a substitute for the lost detail. Depends on the image, the application, and personal taste..

One last thought. Downsampling and JPEG compression are very different processes, it's not particularly helpful to compare the two. But it is worth being aware that if you heavily downsample a large JPEG you tend to disguise the compression artefacts. Another point in favour of downsampling on the PC. The counter-argument is that (unless you are shooting RAW and saving in a non-lossy format), the image has suffered two rounds of JPEG compression..

So as you see, while there is no single categorically-correct answer, it leans heavily in favour of downsampling on the PC..

(*) Actually, there are rare exceptions to this but they don't impact on this discussion...

Comment #4

Thanks for the help, Steve.Elements does have bicubic and bilinear choices for downsampling...

Comment #5

Antony John wrote:.

Thanks for the help, Steve.Elements does have bicubic and bilinear choices for downsampling..

Ok, then use Bicubic. Bilinear is a computationally simpler method which is a hangover from when computers were less powerful. I don't know of any situation where it is useful..

You probably also have "nearest neighbour" downsampling - that one is not suitable for continuous tone (i.e. photographic) images, but it is good for line art (logos, cartoons, etc)...

Comment #6

When you resize to 640 x 480 be sure to add to or make a new file name or add an E (edited) or W (for web) to the file name....this will creat a new file and won't change your original...also use #10 or 12 for jpeg compression which should give you a clean jpeg with no apparent artifacts......Bill,Jr'I kind of like the Earth, it's where I keep all my Stuff.'Website; http://www.pbase.com/wboth125 Lake Wylie, SC..

Comment #7

Hi AJ,.

For your info, both PS and PSE have had Bicubic resampling available (since forever) for use when resizing images, and it's the only real choice for photo work..

What Steve was actually referring to was one of a pair of *variants* on the Bicubic algorithm, called "Bicubic Sharper" and "Bicubic Smoother". These are relatively recent additions, designed to account for the losses inevitably incurred during the resampling process. "Sharper" is intended for use with reductions, "smoother" with upsampling. Both of these are now in place in the latest versions of both PS and Elements (PS may have got it first... not sure)..

The amount of sharpening applied is arbitrary, however (according to the taste of someone at Adobe), and in practice it's impossible to expect a "one size fits all" to always satisfy. I agree with Steve that you're better off just using "Bicubic" alone, and then making your own decision on just how much secondary sharpening to apply after resampling. I find "Bicubic Sharper" a tad heavy handed, for example, when doing Web preparation even though that's it's intended use. (Results will be in part dependent, of course, on how much primary sharpening you've already done in your initial post-processing.).

Antony John wrote:.

Thanks for the help, Steve.Elements does have bicubic and bilinear choices for downsampling..

There's something I should throw in on the issue of jaggies and apologies if you're already au fait with this when you're experimenting with downsampling images destined for e-mail, and trying to decide how far to go in balancing size with quality, for example..

Don't EVER attempt to use screen zoom in Photoshop (or other similar editor) in the hope of getting a useful preview of how a reduced size image will look, subjectively. Such a preview will in most cases look terrible and will be, in effect, useless to you for this purpose. You must actually execute the resample in order to get any realistic idea of how the result will look. The only exception is when screen zoom is an exact, even, sub-multiple of 100%, i.e. 50%, 25% or 12.5%. Forget any other number, since the temporary resampling done just for display purposes is extremely unsophisticated in the interests of speed..

The "Save for Web" utility is quite useful for this sort of work, on the other hand, since the actual resample is done every time (that's why it's slow). Be aware, though, that this route strips away the EXIF data from the resulting image file a contentious state of affairs among digital photographers, that can be either a killer or a non-event depending on where the image is destined for use..

MikeMelbourne.

Image control:Zoom outZoom 100%Zoom inExpand AllOpen in new window..

Comment #8

Mike Fitzgerald wrote:.

What Steve was actually referring to was one of a pair of *variants*on the Bicubic algorithm, called "Bicubic Sharper" and "BicubicSmoother"..

I was, but in addition I wasn't sure what was available in PSE..

I agree with Stevethat you're better off just using "Bicubic" alone, and then makingyour own decision on just how much secondary sharpening to applyafter resampling..

On the other hand I do use Bicubic Smoother for upsampling and it is *much* better. I didn't mention this before as the question was specifically about downsampling..

Don't EVER attempt to use screen zoom in Photoshop (or other similareditor) in the hope of getting a useful preview of how a reduced sizeimage will look, subjectively. Such a preview will in most cases lookterrible and will be, in effect, useless to you for this purpose. Youmust actually execute the resample in order to get any realistic ideaof how the result will look. The only exception is when screen zoomis an exact, even, sub-multiple of 100%, i.e. 50%, 25% or 12.5%.Forget any other number, since the temporary resampling done just fordisplay purposes is extremely unsophisticated in the interests ofspeed..

Even with the exact divisions, it doesn't work. Photoshops's 50%, 25% etc screen views are very soft and just not reliable. Irfanview, by contrast, seems to process it's screen images for best appearance and applies some sort of sharpening to them. I think this is because of Irfanview's focus as a viewer rather than an editor..

The "Save for Web" utility is quite useful for this sort of work, onthe other hand, since the actual resample is done every time (that'swhy it's slow). Be aware, though, that this route strips away theEXIF data from the resulting image file a contentious state ofaffairs among digital photographers, that can be either a killer or anon-event depending on where the image is destined for use..

Save for Web seems to use a different compression algorithm that works especially well for low resolution images. It has an amazing ability to produce very small files which still look ok. But it is for producing web page content rather than photographs as such - i.e. the 25 to 50 kB images we would use for editorial material, product shots, and so on, not the 150 to 300 kB images we use to display our work. The photographers who complain about it haven't understood what it is for...

Comment #9

Click Here to View All...

Sponsored Amazon Deals:

1. Get big savings on Amazon warehouse deals.
2. Save up to 70% on Amazon Products.


This question was taken from a support group/message board and re-posted here so others can learn from it.

 

Categories: Home | Diet & Weight Management | Vitamins & Supplements | Herbs & Cleansing |

Sexual Health | Medifast Support | Nutrisystem Support | Medifast Questions |

Web Hosting | Web Hosts | Website Hosting | Hosting |

Web Hosting | GoDaddy | Digital Cameras | Best WebHosts |

Web Hosting FAQ | Web Hosts FAQ | Hosting FAQ | Hosting Group |

Hosting Questions | Camera Tips | Best Cameras To Buy | Best Cameras This Year |

Camera Q-A | Digital Cameras Q-A | Camera Forum | Nov 2010 - Cameras |

Oct 2010 - Cameras | Oct 2010 - DSLRs | Oct 2010 - Camera Tips | Sep 2010 - Cameras |

Sep 2010 - DSLRS | Sep 2010 - Camera Tips | Aug 2010 - Cameras | Aug 2010 - DSLR Tips |

Aug 2010 - Camera Tips | July 2010 - Cameras | July 2010 - Nikon Cameras | July 2010 - Canon Cameras |

July 2010 - Pentax Cameras | Medifast Recipes | Medifast Recipes Tips | Medifast Recipes Strategies |

Medifast Recipes Experiences | Medifast Recipes Group | Medifast Recipes Forum | Medifast Support Strategies |

Medifast Support Experiences |

 

(C) Copyright 2010 All rights reserved.