Huawei P10 review
Apologies for what may be a Dim Noddy's question, but what little I know about website design dates from five or six years ago.
I remember reading in lots of places that it's pointless to include images with resolutions better than 72dpi "...because monitors can't handle anything finer." At that resolution I'm finding it difficult to produce crisp images of scanned pages of 8pt or 10pt text and I'm having to resort to doing things via OCR; this adds a boring and labour-intensive proofreading/correction operation. The cost of the additional work is not an issue (my website is a strictly non-commercial labour of love) but the project involves a lot of pages, the work is sheer drudgery, and the result on the website no longer looks authentic.
At present I'm scanning with an Epson Perfection 3490 to produce JPEGs at 72ppi, then punching up both brightness (a bit) and contrast (a lot)with Photoshop 7, and finally applying the Sharpen filter. The result might best be described as the best of a long series of failed attempts (including scanning to produce 200ppi TIFs and converting to JPEG only at the Save for Web... stage).
Could someone please suggest a better approach?
Also — in passing — five or six years ago, 'monitor' meant 'CRT monitor'. Can today's TFT monitors do any better, or does the 72dpi rule of thumb still apply today?
Thanks for responding. I try to keep each page image below 30kB and chose .gif because I found that 'outputting' (dreadful word) in .jpg format increased the file size without improving the quality. I thought .gif's limited colour range wouldn't matter since I'm scanning black-on-white text only (i.e. no colours). I have grayscale selected and, curiously, I've found that selecting 3 'colours' gives a slightly better result than selecting 4.
I've been suspecting that the root of the problem may be the combination of small-size sanserif text (8 & 10 point, some of it not terribly well printed) with 72ppi resolution. Am I on the wrong track?
Standard screen DPI is 96, Jim. Any good image editor will default to that while allowing custom settings as well. For example, I often have to use 600dpi for images that will be used in print publishing.
If you have good editing software, you can fine tune the percentage and sub-sampling method of Jpeg compression to produce good results. I can usually get an original high-res c.3500 pixel 3.5MB image down to an 800 x 600 size at c.100KB or less with no discernable artifacts at all. A less detailed image could go to half that size with the same dimensions.
I've had another go, playing with the settings in Photoshop as you suggested. The smallest jpeg I can produce from a typical scanned page is 73k and the artifacts are noticeable. If I optimise the same page as a gif, however, I get a file that's only 35k and artifacts that are less obtrusive.
The material I'm scanning consists of text pages from copies of my old school's magazines. In general each page is well filled, tightly set (9 or 10pt set solid, at a guess) and not terribly well printed in places.
My 'clients' include a number of elderly people who were at the school in the 1930s, 40s and 50s. Many of these folk are on dial-up connections, which is why I try to keep my file sizes as small as possible.
The magazine's page size is 6" x 9" and I wish to reproduce each page 'as-is' on the website. Am I simply asking too much by trying to keep each file below 30k? If so, could you please suggest a more realistic file size?
Regrettably, I have to go out this evening and will be offline until tomorrow morning.
File sizes of that size can be pretty jaggy and there's no real cure for that. There is a point at which no lossy file format is ever going to give great results and at that point, you have to balance to the two opposing needs of file size versus useable quality. Not easy. As you have found, Gifs are smaller, suffer fewer artifacts and are quite well suited to simple reproduction suh as text. The trade off is the very restrictive colour depth in images.
It certainly seems that any solution is going to be a compromise, just as you say. I'll have to accept that viewers on dial-up aren't going to get crisp, sharp text and 'instant' downloads.
Two more questions if I may: you wrote "If you have good editing software, you can fine tune the percentage and sub-sampling method of Jpeg compression to produce good results."
1. Am I correct to assume that the percentage adjustment = the Quality slider in Photoshop 7?
2. What's the sub-sampling method called in Photoshop 7?
Thanks for the info. I'll tick the box now.
Hi Jim. I don't use Photoshop - I use Photoimpact, which does almost as much for 1/6 of the cost! - so I'm not familiar with the details of it. If the slider is like mine, it will have figures on representing the percent values. There may also be some default steps that you can use, although fine tuning is often better. The sub-sampling methods are likely to be None, YUV411 and YUV422. 422 makes bigger files, None makes smaller files but the quality is poorer. Then there's standard and progressive mode. Progressive drops the file size a little bit more.
Posted too quickly - I was going to mention that the compression options should give you a readout of download time on a given connection speed. I can slide mine from 14.4k to 512k. I wish it went up to 8mb, but a bit of basic maths works that out! I leave it 56k.
Many thanks, ade.h.
Those YUV designations don't appear anywhere in Photoshop 7 as far as I can see, but I'll do a little Googling and see how I get on. Photoshop 7 does indeed give an estimate of download time for a given connection speed, with a stepped scale from 9.6k modem(!) to 2M but I reckon it has to be indicative only, with numbers that are a touch optimistic in relation to the real world.
I leave mine at 28.8KBps on the assumption that I'm balancing out the optimism.
This thread is now locked and can not be replied to.