Microsoft's HD Photo Format at a Glance

I have put together a short rundown of how the Adobe Photoshop beta plug-in for Microsoft's new photo format performs against current formats. I ran the plug-in through a very basic set of tests. The aim is to get you thinking about the results. Enjoy!

I started off by taking a screenshot of Neowin's header and cropping it in Adobe Photoshop CS2. I then saved it as a Portable Network Graphics (.PNG) file, a Joint Photographic Experts Group (.JPEG/.JPG/JPE) file and finally a HD Photo File Format (.WDP/.HDP) file.

To my surprise, Photoshop presented me with a fairly extensive settings window for my WDP file. The objective of my first test was to keep everything on default settings, so I hit OK, promising myself I'd explore the available options later. Photoshop saved both the .JPEG and the .WDP files as copies, meaning I had to open the files up again.

I headed over to ImageReady, only to have my monitor happily throw out an error. That's beta for you.

Undeterred, I decided Internet Explorer 7 would save the day. It didn't. Neither did Paint.net. No, I did not bother with Microsoft Paint. On the bright side, Vista's Windows Photo Gallery opened all three and naturally, Photoshop did as well. With the former, I used the arrow keys to move between all three images, but I failed to notice any discrepancies. Since this wasn't a very accurate test because the image wasn't very complicated, I decided to move on. I did, however, note the file sizes. At default settings, PNG, JPG and WDP gave file sizes of 38.6KB, 34.4KB and 36.6KB, respectively.

Next, I used a 58.0KB PNG file with an intimidating number of colours and re-saved the file as JPEG and WDP files. Once again, not exactly the best way of comparing the remaining two, but at least the source was constant, and I could compare the file sizes. I used maximum quality settings and found JPG at 2.67MB and WDP at 8.74MB (lossless). That's more than triple the size difference. It is lossless after all. When it comes right down to it though, chances are the spectrum will blind you before you can find a difference. Comparing all three against each other in Windows Photo Gallery gave me nothing the human eye could detect. Nonetheless, it was comforting to see that the option for lossless was present.

I decided it was time to go for the lowest file size possible. Since this is all about Microsoft's format, I went ahead and grabbed one of many Microsoft logos off their support website. The results of this test, wasn't something I was expecting. At the cost of quality, the plug-in allowed me to get a smaller file size than .JPG did, 19.6KB compared to 24.9KB. The quality was in favour of JPG by a lot. And yet, when I saved the file as a PNG, it managed to keep all the aspects of the image, at a mere 5.35KB. That's right, it would seem that PNG is the clear winner. Notice that in the image below, the images are sorted by size.

Finally, I took a JPG of Bill Gates (33.4KB) and continued saving to WDP while setting the quality bar lower and lower, until I reached a smaller size than JPG (29.0KB). This was when the slider was at 0.5. Changing it to 0.4 basically corrupted the image. So tell me, can you see the difference?

In conclusion, it appears that WDP is much more flexible than JPG is: from lossy to lossless. At least when it comes to Photoshop, it has many more options available. That doesn't mean it is necessarily the superior format, but at least it has the edge in the features department. It does still have a long way to go though.

If you happen to have CS2 or CS3 installed, go out and get a copy of the beta plug-in. For any wallpaper creators out there, and I know Neowin has its fair share, it might be a good idea to compare how your work is displayed in Microsoft's format, against every format available. Run it through your own tests, and report back. Comment on whatever you find!

Note: The gallery for this post with all the file types is currently unavailable, mainly due to the fact that it only takes PNGs and JPGs When I get it up again, I will include many more comparison shots.

View: Gallery
Link: More Information on HD Photo

Report a problem with article
Previous Story

Microsoft: DoT Vista snub won't ripple

Next Story

Turkey lifts YouTube ban after two days

53 Comments

Commenting is disabled on this article.

Hm.....
you can see difference only if you compress picture to the maximum, and who will make that ? If you converting picture you like to be good not to be distorted. When converting from RAW or PSD there is not difference. I take shot with my Canon 400D with RAW quality and compare pictures. There is not BIG differences. It is hard to compare all formats because WDP format can be opened an seen only in Windows Photo Gallery in Vista and Photoshop. PSD format can't be seen in Windows Photo Gallery so you can't compare this formats here. Programs like ACDSee or XnView have support for PSD, JPG, JPEG2000 but not for WDP, so I compare all formats but screenshot is only from WDP 16 bit at Low quality and JPG 8 bit at Low quality saved with Photoshop (0 quality). It's a part of a picture. Test archive with all formats is 35,0 MB I can't post it here.
This is RapidShare link: Fileformats

How can I post image here?

I hate to join the dogpile, but the author really should've done some research and thought things over before going about his tests... or, at least, publishing them. The format is called "HD Photo." What about that is all that hard to understand? Not even one of the tests in the article dealt with a hi-def photo. As has been suggested countless times, take a large (at least 1600x1200) RAW or TIFF image and do the comparisons from that. As one poster did, you get about half the file size of JPG, which is--suprise--exactly what Microsoft claims.

Neowin should know better than to do such inaccurate testing. This is something that I'd expect on a Mac zealot site to trash on Microsoft.

Finally, I took a JPG of Bill Gates (33.4KB) and continued saving to WDP while setting the quality bar lower and lower, until I reached a smaller size than JPG (29.0KB). This was when the slider was at 0.5. Changing it to 0.4 basically corrupted the image. So tell me, can you see the difference?

It isnt clear from the description but it sounds like you took a jpeg and then began saving in WDP until you got the file smaller. If this is the case theres no conceivable way you could have gotten a better image than the original Jpeg. If your going to do comparissons the source file should be a lossless one.

rm20010 said,
Install .NET Framework 3.0

No, installing .net 3.0 won't do a thing. You need either Vista Photo Gallery or Adobe Photoshop to open the Wdp pics.

I agree with NickJ, saving a jpg image that's already had artifacts introduced as a wdp image isn't a valid test, you have to start with an uncompressed master image. It's impossible for the WDP to be as clear as the jpg unless you had saved it lossless, which would have been pointless for this particular test.

Sorry mate don't really rate your test! You should have started with uncompressed TIFFs taken @ like 12mp with a digital SLR, then compressed them to a set of differing filesizes. Small screenshots and logos show nothing, and recompressing a compressed source is useless.

Ehh? Aren't the images that you posted up here compressed? O_o

Makes for a rather unbalanced argument unless the differences are as plain as night and day.

You do realize HD Photo is better than jpeg2000 simply because of licensing options alone, right? Also, the test posted in the article is not well done and therefore not indicative of actual results. Look a few posts up. Same image quality, yet the HD Photo version is 45% smaller in file size. Now that's damn impressive.

You shouldn't really compare HD Photo to JPEG, but e.g. JPEG 2000. Everyone knows HD Photo would come out as much better than JPEG; it's like comparing RAR or 7zip to zip, or mp3 to Ogg or AAC. Two different generations! Microsoft isn't lying when they're saying it's better than JPEG. One problem with adoption here may be that we're not as restricted to small sizes as we were in the past with both smaller drives and lower bandwidths. I can't really say I'm feeling an urge to switch, and both HD Photo and JPEG 2000 may fall into the same situation as the competitors to mp3. Sure, they're used, but not even nearly overtaking mp3 use.

Anyway, here's to me a more interesting comparison between HD Photo and JPEG 2000:
http://www.compression.ru/video/codec_comp...parison_en.html

The full test document here as a 4 MB PDF:
http://www.compression.ru/video/codec_comp...mparison_en.pdf

Graph with codec comparisons below. The bolded lines are HD Photo codecs, the rest are JPEG 2000 codecs. Higher means a lower average per-pixel difference from the original. So in this case, ACDSee's JPEG 2000 implementation came out on top.

[img]http://www.libyano.com/HDP.wdp[/img]

Both are saved in 90% quality
The source is a RAW image taken with Canon EOS 300D

As you can see the JPG is almost 2 times bigger than the HD.

I am just a regular user and no one pays me.

Here's a JPG 2K encoding too just for completeness and comparison: :)
http://northgrove.googlepages.com/jpg2k.jp2

Of course, that one is recompressing a JPG source though, so it may struggle a bit with some minor artifacts introduced that is now adding complexity to the image. The codec implementation used was that in the free application FastStone Image Viewer.

I think you've missed the whole point of the format. Now I only remember a very small bit about it form one of the original new postings about the format from a week ago or so, but wasn't the new Microsoft HD format supposed to handle many times more colours than .jpg?

In the news posting http://www.neowin.net/index.php?act=view&id=38641 it says that it can handle from 8 to 24bits more per pixel than .jpg which is stuck at 8bits per pixel (maybe jpeg2000 and some newer formats are not, I don't know). Now maybe its just me but I fail to see how you can compare something that was created at 8bits per pixel (for those images you compared that came from .jpg origins, or from screen shots of the website stuff). You can't create more information than is available (or you can and it's crap/false/fake data).

I think the idea would be to take a large RAW file or .tiff image (that had that format as its origin) and do compression comparisons from there. Then you can be confidant that you're results showcase the actual abilities of the various formats in the comparison. As it is you're comparing a crippled format (.jpg) with re-compressions of itself.

Frazer

Nice guide, but I think I will wait till the beta plugin is final before using it for any production work.

Least it's shows me what MHD format is all about though.

For the gates picture the .jpg is better quality, it's crisper, the .wdp has a blur to it, although colours match.

What would be a test that I would of expected would of been to get a wallpaper, save a jpg to say 80% qual, which is rather accepted, and then try and get the .wdp image to match the filesize of the compressed jpg, and then see which one has better quality...

The HD Photo as well as JPEG 2000 standards supports 100% lossless encoding modes.

According to Microsoft, HD Photo can compress an uncompressed image lossleslly about 2.5x.

GP007 said,
You do understand that PNG is not for Photos right?

And an iPod isn't supposed to be a Black Box for an airplane, but they're doing it. When I have the option of either having a 300MB TIFF image of a galaxy and a 60MB PNG image of the same galaxy, I'll take the PNG image since it gives the exact same quality.

Wow, I think I'm going to continue using PNG24 instead of everything else. I already didn't use the JPG format anymore. Anyway, if it's better than JPG and it's still in beta...

It's not fair to compare a jpeg with another format that's a lossy compression of the jpeg, because jpeg has lots of artifacts of compression in the first place. The human eyes may be insensitive to the artifacts but they are definitely noises to the original image. Noises are difficult to compress. A better comparison is to take a raw image then convert it to both formats of roughly the same size. Generally there're many many tricks (filters) that can be used to make a image more compressable, mainly by removing details that are insensitive to human eyes. Also it would be helpful to compare different types of images (outdoor/indoor, high/low contrast, scenery/human face), and zoom to reveal details of the results.

The comparison of the microsoft logos is is misleading, that is *exactly* the sort of image png is designed to for! (large blocks of same colour) and exactly the opposite of what photo formats are for!

I'm always amazed at how many people do not realise this.

On the Bill Gates pic, if you zoom in a bit, you can see his cheek has a little more detail in the left picture. It seems to make the picture smaller by 'smoothing' it, by removing minor details in the picture. This is definitely nothing revolutionary, other formats including JPEG2000 and JPM use similar methods of compression. It's nothing you'll notice with basic use, but if for example you were to get a large print of the picture, then you would probably notice the loss of detail.

I bet if we were provided with a larger copy of the Bill Gates picture, we would definitely notice some differences...The one in the article is too small for us to judge.

Hm... Professionals should do the tests, you must know the best settings for both you get small file with very good quality.

It's the same like with video. You can make 8GB and 4GB HDTV encode and the quality could be the same... Everything depends on your settings...

why didn't you test it with a normal picture (humans with a background) and make a wdp the same size as a jpeg. That would have been interesting. A two tone picture (the logo) proves nothing, ofcourse png is the smallest, it resembles gif, it's made for this kind of things. A real picture would have been interesting, this means nothing, sorry

I tried it out in with a 1.99MB image (a 1600x1200 jpg), at it's default setting it produced an image without any noticable difference at 1.37MB, lossless mode gave me a 3.35MB image and the lowest setting gave a horrible pixilated image at 29KB.
I managed to get the image to 165KB before I started noticing a lot of artifacts.

well i can't see any difference and the hd photo size is a quarter less, so that's a plus for hd photo i guess :), we should have the original in high quality png or even bmp to be able to see if there are any details left out

Well, with the naked eye I can see that the jacket on the JPG picture is a bit more fine detailed, and I can also note the hairs behind his ears a little blurrier on the WDP image as well as some weird color differences between the JPG image and the WDP around the back of his hand.

Overall, no true noticeable difference unless your job counts on 100% precision.

Ghostdraconi said,
I tried it out in with a 1.99MB image (a 1600x1200 jpg), at it's default setting it produced an image without any noticable difference at 1.37MB, lossless mode gave me a 3.35MB image and the lowest setting gave a horrible pixilated image at 29KB.
I managed to get the image to 165KB before I started noticing a lot of artifacts.

I say that's pretty good for a new format with a beta plug-in. We should wait for adobe to get a newer/final plug-in out (or whoever made it to), and then run the test again.

Slimy said,
I will save that for the gallery. For now, tell me what you think about the Gates picture.
I think Gates has aged considerably :ponder:

The thing is, it's hard to measure the quality of an image. I did the max (lossless) and the min (quality = 0) but that's just the beginning. It will get even more interesting if you can get the exact same image at a smaller size (which is what Microsoft claims). But how do you know the smaller file size image really is the same quality as the bigger one?

Not really. These tests weren't very fair except for maybe the last one.

This is a photo format. It's not meant for logos. The compression scheme isn't supposed to work with them, as you can see by how it butchered the simple MS logo.

The _best_ test would be taking a huge RAW image, and compressing it.