New limits of my Epson 4490

[carlos_rodriguez3]

<p>I have read tons of threads on scanning. Most of them suggest Epson 4490 and V500 users to scan at 2400 dpi. The fact that those scanners are not dedicated film scanners (have a glass, not perfect focus), limits their true scanning capability to 2400 or less dpi.</p>

<p>But then I got curious. I found an almost perfect slide with a fence that goes to the distance at the limit of my scanner sensitivity. In my original scan at 2400 dpi, the fence (at the end) looks smudgy and undefined. Epson claims the scanner optical dpi is 4800 so I did a small test. And what I found was surprising: the fence looked better, not perfect, but better.<br>

I reduced the picture to 50% and no detail went missing (at least to my eyes). So I concluded that the true dpi of the scanner was 2400 dpi, BUT (big but) you need to use 4800 dpi to obtain it. I am not sure why, but I made several test with different sharpening and curves and found that almost all of them the 4800 dpi version was better. I used a high contrast of the picture in magnify the details.</p>

<p>This contradicts everything I have read on this scanner. I dont know, maybe I am missing something, but this is what I found. I included a comparison version of the same portion of the fence (the critical part). And I also included the complete picture. What do you think about this?</p>

<p>Just think you would be interested</p>

<p>Carlos Rodriguez</p>

<div></div>

[carlos_rodriguez3]

<p>I forgot to tell you the slide is a Kodak Elite Chrome 100. Here is the original slide:</p><div></div>

[larrydressler]

<p>Adjusting the height of film holders also works. those little sticky felt pads can make a difference.</p>

[marc_bergman1]

<p>Carlos,</p>

<p>I found the same thing with my 4490. It is not noticeable with all slides but for some it does make a difference.</p>

<p>That is why it is important to do your own testing. There are going to be sample variations in all products.</p>

<p>I don't use it for 35mm since I got my Nikon but may try it with some of my medium format slides. It does produce a huge file at 4800.</p>

[thirteenthumbs]

<blockquote>

<p>Most of them suggest Epson 4490 and V500 users to scan at 2400 dpi</p>

</blockquote>

<p>That is based on what that person can see with their eyes. Instruments can detect more than the eye and the results one gets is dependent on the test procedure also.</p>

<p>I scan at the manufacturers stated optical limit, do my dust/scratch removal, adjust the tonality to suit me, then save that as my master file. </p>

<p>If independent testing labs using the same quality level of test target, monitoring equipment, and test procedure were to get different results than the manufacturers I would possibly believe them. I take magazine reviews and store test results with the appropriate grain of salt. </p>

[peterbcarter]

<p>I have owned a 4490 for about 3 years now and I have not seen any benefit of going past 2400. I have done basic tests (scan higher until no difference in quality is apparent). I have a Plustek 7200 (which gives an estimated scan of 3200 dpi). There is much more detail from the Plustek. Perhaps you are seeing more detail than 2400, but that doesn't mean you are seeing 4800. The accepted amount (opinions around the net) is around 1800.</p>

<p>I use vuescan (mostly) and silverfast ai (when I want to torture myself) with the above statement being true. You did not mention what software you were using, how you were saving the file (or acquiring it directly into a program). Knowing this can make a difference when comparing my experience to yours.</p>

<p>What I think you have seen is a sweet spot where the dpi is different among the X and Y axis. I suggest you remember the orientation and take advantage the next time you have verticals. If it is a mounted slide, you can just put it on the glass with the opposite orientation and do another sample. I suspect you will get different results.</p>

<p>Assuming what I have said is completely true, it is noteworthy and can be taken advantage of when you have an image with lines of detail. Another 'trick' to get more detail out of a perfectly useful scanner.</p>

 

[starvy]

<p>I found 4800 dpi to take too long for the strip of 6 35mm frames or four slides. I actually tend to do 2400 if I think they are worth putting up on my facebook library or 1200 as tests to store in computer.</p>

[carlos_rodriguez3]

<p>To Peter Carter:<br>

I used Epson software, provided by the product. I am using TIFFs.</p>

<p>I will put the slide sideways and test it that way. I let you know what I got</p>

[peterbcarter]

<p>Yes, I would like to see what comes out of it.</p>

[carlos_rodriguez3]

<p>Peter:</p>

<p>I scanned the slide vertically. It seems you were right. At 2400dpi there is almost nothing there. But still at 4800 dpi rhere is some detail and some lines. It is different, but much less defined.</p>

<p>I am at the office and the scanner is here, but I am using a laptop. Once I have acces to my house PC, wchich has a large screen, I will post the examples. This will be done this night</p>

[StuartMoxham]

<p>When the wind blows in the right direction I can sometimes get more detail by setting my V500 to 6400 ppi but other times it makes little if any difference. At any lower resolutions the car grills would just blend to a solid mass..</p><div></div>

[art_thomas1]

<p>I have an Edmond Scientific glass test target with the lines made from chrome electro deposited on the glass. It is used for measuring the resolution of microscopes. Over the last few years I've tried it out on a number of scanners. I like my Epsons and use them quite a lot, but do not expect more than about 2500PPI out of any of them. <br /><br />When using a test target made up of lines various spaces apart, there is a judgment factor in determining at what point one should call it the max resolution. There is a German web site called Scandig that tests scanners. Since I get about the same results I am at least making the same judgment as to where the separation point is. Two other things...<br /><br />Registration/alignment. I do 10 tests. I reposition the test glass each time. If a test line is not exactly in line but a little off angle with the scanner path it can cause a lower score. Also, when the test lines get to the same fineness as the scanners capability the scan photosite must be centered exactly over the line. If not, the results will be artificially low. <br /><br />Grouping. The test target does not have a value for each lp/mm number. Perhaps one set of lines would indicate 66 lp/mm and the next finer would be 77 lp/mm. I might see the 66 as absolutely clear and the 77 a little worse than perfect and pick a number such as 70 for a value. You might pick 69 or 71. One person might say 66 was the last perfect so that is "my" score. Another might say "77 is not perfect but close enough so I'll value it at 77. Who would be right? <br /><br />It has been my experience that the Epsons which claim 4800-6400PPI all test out around 2300-2600PPI with the difference not between models but between samples.</p>

[carlos_rodriguez3]

<p>Art:</p>

<p>Very interesting points. However, my initial claim doesnt dispute the fact that epson scanners cannt scan beyond 2400-2600 dpi, but that you need to put the scanner software at 4800 dpi in order to get 2400 dpi numbers.</p>

<p>Peter:</p>

<p>Here are the results for the fence picture horizontally. As I commented, the results are much less impressive. The 2400 version just shows a "ghost" fence, but the 4800 dpi version shows some fence. It is very faint, but I see it there.</p>

<p>And I am still intrigued and excited with this discovery</p>

 

[carlos_rodriguez3]

<p>Art:</p>

<p>Very interesting points. However, my initial claim doesnt dispute the fact that epson scanners cannt scan beyond 2400-2600 dpi, but that you need to put the scanner software at 4800 dpi in order to get 2400 dpi numbers.</p>

<p>Peter:</p>

<p>Here are the results for the fence picture horizontally. I scanned the slide vertically, so the fence would be horizontal to the front of the scanner. As I commented, the results are much less impressive. The 2400 version just shows a "ghost" fence, but the 4800 dpi version shows some fence. It is very faint, but I see it there.</p>

<p>And I am still intrigued and excited with this discovery</p>

<div></div>

[andylynn]

<p>Okay, here's the deal with these. The actual amount of information you get in the image file is going to be limited to less than 2000 PPI. Probably more like 1600. BUT, you will not get that much detail if you set the scanner to 2000 or 1600 - to get that much detail, you need to set the scanner to max.</p>

<p>I have a V500 and I've done comparisons between scans at 1600 and at 6400, downsampled to 25% - same number of pixels in the end, but the downsampled version is much more detailed.</p>

[art_thomas1]

<p>" but that you need to put the scanner software at 4800 dpi in order to get 2400 dpi numbers." </p>

<p> **********</p>

<p>Quite true. I believe it is because although there are enough photosites to make 4800PPI, the optics are not good enough to transfer that to the final in machine receptor. You get a reduced percentage of resolution, about 50%. Ergo, if you start with 2400, and the optics transmit less, your resolution is again a percentage of the photosites being used. Since poor optics are non linier as to their throughput, you would probably get 60% rather than 50% of the 2400 figure or closer to 1500PPI equivalent. </p>

 

[mwmcbroom]

<p>I own an Epson 4990, which is very similar to the 4490 in terms of resolution capabilities. So I read this topic with great interest a couple days ago, and tested my 4990 yesterday. I didn't have any negatives handy with fine grid lines. Instead I just scanned an old negative portrait of my daughter, thinking I should be able to spot some differences. I scanned the negative first at 2400 ppi, and second at 4800 ppi. Then I compared the two -- first with the 4800 ppi scan enlarged by 50% compared to a 100% enlargement of the 2400 ppi scan, and then with the 4800 ppi scan reduced in image size to the same pixel size as the 2400 ppi scan.</p>

<p>Well, I had just about given up on trying to spot differences between a 2400 ppi scan and a 4800 ppi one, when I spotted it. Smooth curves were smooth curves on the 4800 ppi scan, whereas I could see the zig-zag stair-steps along these same curves with the 2400 ppi scan. My daughter's cheek line was smooth on the higher res one, whereas the stair-steps were subtle but noticeable with the 2400 ppi one.</p>

<p>So . . . this was enough for me. Any tiniest bit of detail I can extract beyond where it seems the limits are, I will try to extract. I've begun scanning a stack of old photographic negatives and I'm looking forward to a good subject where I can give it a more rigorous test than just another pretty face. :)</p>

<p>If/when I run across a negative with a suitable subject where I can demonstrate a difference, I'll come back here and post the results.</p>

<p> </p>

[larrydressler]

<p>About time people believe what I have already known. :-)</p>

[peterbcarter]

<p>Michael, the 4990 is more of a V700 unit. Not quite the same light hood (no 8x10 and a regular lamp vs led), but almost the same guts. Your results are no surprise, the unit has a better lens and far greater DMAX (3.2 vs 4.0).</p>

<p>In my opinion, the biggest reasons to step up from a 4490 to a 4990 (or V7x0) would be the increased DMAX, larger film size and a better lens. Especially if you do chromes,</p>

[david_love2]

<p>Larry: I believe that you use a v700. Do you use the max resolution setting on it, as in the discussion above? Also, do you use the original film holders, or an after-market set. I appreciate your knowledge in this forum and look forward to your reply.</p>

[larrydressler]

<p>Yes I do use the V700. I use both the stock holders and the AN glass depending on the negative. Many lay totally flat so I don't want to add another layer with the glass. Yes I found that scanning At a larger size and then down sizing gives more detail. However for most work I don't have to do this and won't because of time. When I am going to make a printed enlargement I do it.<br>

With MF there is almost no need to scan that large.</p>

<p> Larry</p>