|
|
|
|
Before I start - do not expect this to be another user manual - this review is about some of the many advanced functions - those which I personally rate as having an important and very visible influence on the outcome, the digital file, which should be the best possible digitization of your film based image.
SilverFast Ai Studio - Selected Advanced Functions
Scanning is the first and in my opinion all-deciding step in the semi-digital workflow of photography (I do not talk about scanning of printed targets as this is in my opinion a different pair of shoes). Semi-digital because your original master is still a film positive or negative and by scanning it, you make it suitable for the digital color workflow, which is indispensable in todays photographic world. And with scanning, the problem starts. Most people are not aware what all can go wrong or what all you can already do during the process of scanning. Most scanning programs, which come together with the hardware from the manufacturer of scanners are good enough for very basic scans and are usually not optimized for scanning of transparent masters. For that reason, you usually get a free copy of the probably most widely spread scanning program, SilverFast SE, packed together with the scanner and its operations software. But that version of SilverFast is for most scanners the Lite version, not the full version. The full version comes usually at a much higher price when you buy the Pro set of a scanner. I would like to review some selected advanced function of the full version of SilverFast Ai Studio, which is the complete program with all features. The scanner for this review was the Epson 4870 Photo as well as the Nikon Super Coolscan ED8000. For certain features I compared it to VUESCAN, which is a powerful yet inexpensive and rather easier to use program. For more details, please go to my review at this website: VUESCAN Let me start with a very basic fact the price for the program. SilverFast is expensive, very expensive if you compare it to VUESCAN. It costs about 4-6times as much and is working with one scanner only the one you bought it for. Yes, one can say, why would I need the program to work with more than one scanner, but think about new models, other brands you might be tempted to choose next time you buy a new scanner or think about two scanners one flatbed scanner and one dedicated film scanner. With SilverFast you need to buy a second license and you need to install the program a second time for a second scanner this is not only unpractical, it is really cumbersome as you now need to discern between the two programs which are simultaneously existing on you computer. This was exactly what happened to me I use the Epson 4870 Photo for 4x5 inch and 5x7inch transparent slides/negatives and the Nikon ED 8000 for 120/220 and 35mm film scanning. I had not only to get a second serial number for registration of the program I also had to install the program a second time on the computer. It works, it works fine but still, it is a burden and in my opinion not necessary. OK, I have installed my second SilverFast and now I can work with both programs well, yes but only theoretically. SilverFast was not starting at all; it crashed my Photoshop when I started it up. After several emails forth and back to and from the customer care desk (they were very friendly and helpful), I found it out I had to disconnect my webcam. Strange, but it did not like the USB2 webcam to stay connected. Once I unplugged it, the program started working. Working with SilverFast Ai Studio is possible on two levels of sophistication the standard level and the expert mode. I do not refer to the standard functions as I presume that most of you are familiar with them because they are more or less identical to the ones of SilverFast SE which is the version usually delivered with better scanners in these days. And of course, all these functions work well, fine and give good results. What I was interested in are the special functions, the ones, which are more sophisticated, more complex and less common. I wanted to see if they really work well, if they produce what they promise to do and if so, how do they compare to similar functions in VUESCAN. One of these functions, which are really interesting, is the SilverFast SRD option the software based Scratch and Dust Removal. Dust is one of the very common problems for scanning and we all know how cumbersome it sometimes is to sit concentrated in a session of Photoshop and clone the dust out of the blue sky or the white sand or other unstructured areas of an image. You can imagine how nice it would be to have an automatic de-duster in a program. So, let me start with my experiences with that option. SILVERFAST SRD I have chosen an image, which contains a lot of details as often my images do and sky. Here in Colorado, where you have about 300 sunshine days, blue sky is normal and nothing too special. The location was at the riverbanks of the Arkansas River; the film was a color negative 120 film ( Konica Color Impresa 50). As usual, if you pull a negative out of your file folder, there is some dust visible within seconds. On purpose, I did not clean or treat the negative with an antistatic device, I put it into the scanner, set the film make, name and speed and scanned it. Dust is visible especially in the sky (well maybe not so well on the screen with its reduced resolution, but on the original in Photoshop), the busy other parts of the image hide dust and similar defects pretty well.
No color corrections made, no brightness and contrast adjustments, just press all automatic and go and scan. You can see there is quite a lot of dust in the sky and the picture also contains a lot of contrast differences between the sunny parts and the ones in the shadow. Will be interesting to see how the program deals with these issues. To compare how an uncorrected scan looks like, I started VUESCAN and scanned the image as well. The same adjustments - I entered the film make, name and speed as well and this resulted in a different, color rendition everything else was on automatic as well. Here the scan from VUESCAN, again with dust and other particles scanned as well.
Both times, the scanning parameters were set to obtain about 300dpi resolution at about 8x10inch print size and ICC profile correction was applied as well. For publishing here on the web, I had to compress all images significantly the original image above is about 24MB large and the compressed one here about 200kB a lot of the fine details are lost of course but I still think you can see what I want to show. Back to SilverFast and the SRD option (the name is the abbreviation of Smart Removal of Defects). SRD works on layers you chose defects of a certain size, remove them, take another layer, chose defects of a larger size and remove them and so on for up to four layers. Thats what is written in the manual and I started trying it out. Defects and dust are inconsistencies in an otherwise more or less homogeneous surrounding the program should detect that there is something, which does not fit the pattern of the surrounding areas and mark it as a defect (dust, scratch etc). But in real life, there are lots of details in images and how would a program see the difference between twigs on a tree and a particle of dust in the water? On this medium format negative, I tried the "Automatic" function...
What you see here is the part of the image, which is containing the most visible dust particles the sky. With some adjustments to make, you can mark the dust particles nicely and they will be removed from the scan. But this is only a part of the image if you look with the same settings at other parts, you can see that the algorithm marked almost all details of the water, the little waves, the reflections, basically everything which appears out of place on that location.
All parts, which are marked red, are to be eliminated and replaced by an averaged content based on the surrounding pixels. This is not acceptable and unusable for this type of images. The image lives and dies with the thousands of little sparkles on the surface of the stream, right? Here is that detail again without the markings (no sharpening)
Well, this was not really encouraging at all. I tried for quite a long time to make it work better, but it always boiled down to the same two options either the tiny details in the water are preserved but then the dust remained or the dust went away and the water details were eliminated as well. Doing manual cloning in Photoshop is boring, but I am pretty fast manually and I have the impression that it is faster than playing with this feature to make it work properly. I tried another sample - this time a 35mm Ektachrome slide (a view of the ancient city of Bremgarten in Switzerland) and that how it looks (in Vuescan with the dust manually removed):
The next image shows the three most visible dust particles highlighted with a dark dot:
I tried again the automatic function and no results at all. So I started the manual SRD option - make a mask for the sky and then go for the individual settings of the function
When you start, you need to draw the mask with the lasso tool (you have options for that tool, just click on its logo) inside the small window on the right. I wish there would be a larger window for drawing the mask on the entire image - first you do it in a pretty rough way
and then you can fine-tune the mask in the moveable yellow larger window - which works very fine. For subtracting parts, you need to press the ALT-key with the lasso tool (on a PC) and for adding the SHIFT-key with the lasso tool.
As soon as you have finished the mask making, you can try again the Automatic mode - which now works inside the masked area only, but in my case, it did not work again. I therefore started the manual SRD options again - after a bit of adjusting the detection sensitivity, all dust particles have been marked properly.
Press the OK button and start scanning and - Heureka! - it worked very well - all dust particles had been identified and removed.
After these experiences with 35mm film, I wanted to see how medium format color slide film performs. What I used as film was Fuji Velvia on my ALPA12SWA, equipped with the Schneider 58mm Super Angulon - the shots were taken at the Pawnee Buttes in Northern Colorado, in the middle of the Prairie grassland. Here comes the raw scan - no corrections, only some sharpening, scanned in fully automatic mode - including the dust of course
You can see, there is dust on the slide - I had to put dust on this slide by letting it sit around on the floor for a while - anyway, dust is visible and needs to be removed. The next image shows a screenshot of the SRD option in fully automatic mode - it is interesting to observe that on this target - color positive slide film, which means that by definition, dust particles are black and not white, the recognition works better then on color negative film and is more selective.
You can clearly see that the dust particles are marked almost perfect - so here it seems to work fine. But is the entire image marked properly? Another part of the image are the butts and they are more or less untouched by the dust eliminator
You can see a few false marks and again - the longish dust particle in the center part of the image is only very partially marked. There is a busy feature in the image, a plant and this palm is partially in the sun and has got therefore sunny spots on the leaves - and here we go - again a lot of false markings. Not as many as on color negative film, so clearly better, but still not acceptable
A lot of the sunlit ends of the palm leafs have been automatically taken out after I played a while with the adjustments but if I try to leave the dust marked, these palm leave tips remain marked as well - just for completion - the fully automatic marking did not work too bad, just it marked much more of the palm leafs. What happens now if you say, I don't care, let the automatic do what it wants to do - well, here is the result - it is a detail taken out of the screenshot and you can see what it does to the palm leafs
OK again, back to the SRD menu - no automatic mode, first again masking the sky, then setting the appropriate parameters - as shown before - and everything worked well, all dust particles have been removed and nothing else was harmed. To lighten the dark shadows of the plants, I have used another function, which I shall explain later, AACO - a contrast optimization algorithm. But the main task, removing the dust was achieved.
My conclusion to this SRD function is - do not use the automatic mode, it either does not mark the dust or it marks too much. Take then time and make a mask for those areas, which you want to be cleaned and then try the automatic mode and see what happens - it is just a push of a button. If it is not to your satisfaction, then use the manual mode - this works and it works well once you understand how it works and what its limits are.
SILVERFAST GANE The grain and noise elimination algorithms are software based and designed to eliminate or at least significantly reduce the appearance of film grain or digital noise. Both of them are pretty annoying if too much visible - unless you are in art photography and grain or noise are part of the artistic appearance of an image. I have tried this function with various films and film formats and my overall conclusion is that in general, it works pretty well. Let us start with a rather easy example - again the image of the ancient city of Bremgarten.
Let us take out a detail from about the middle of that image and see how it looks like under significant magnification
The film used was a Kodak Ektachrome 100, so one of the rather small-grain films, but still, in this magnification, you can see a lot of grain - and that the scan was a such a high resolution that the grain is visible. If you reduce the scanning resolution then you would see there instead of the film grain the pixels of the scan. To apply GANE is very easy. You prescan your image and then select GANE as digital filter option in the relevant pull-down menu. You can select one of three intensities of this filter, from light to heavy grain reduction. I have chosen the middle one, medium GANE and the result looks is the following image detail:
One can clearly see that the grain has been suppressed quite a bit but the overall image has softened very much. If you look at the smallest details, they are still there, but this image needs sharpening again - I chose additional sharpening after the GANE filter was applied (I did not find a way to sharpen the image within the SilverFast program after GANE was applied) with Unsharp masking in Photoshop. The result is quiet fine and appealing - don't forget it is a small detail out of a large image what you see below!
Many of you know probably the program "Neat Image", which one could declare as a competitive grain and noise removal program. One of the versions is free and I tried the most recent version, Ver.4, on the same image
I am not certain which approach delivers the better result if any - a direct comparison screen shot maybe is of interest
The color preservation is better in the right image but if you look at the dark areas, there is more noise in them and they are still grainier. A stronger grain reduction would result in even more softness, so I did not apply it in this image. I think that the results justify applying this filter - taking into consideration that on an 8x10 enlargement of this 35mm slide, the area in the screenshot above is maybe not larger than 1inch... The next image for this filter was much more difficult - it was a shot taken on a 1000ASA film, the Kodak PMZ Pro 1000, 120 format special color negative film. Parallel, I shot the same subject on a standard 200ASA film. Here is the complete shot
I filtered the image with GANE in medium intensity and took a part of that filtered image (having removed manually some of the dust particles) - a cross section of a red onion - that part where the onion has got its smallest details and here is that part without and with GANE - quite a difference!
I would say, a pretty good job considering the huge amount of grain in that image (don't overlook the ISO1000/1000ASA film speed). I also tried the more intensive settings of GANE with adjustments possible in the Expert Mode, but the image did not really get much better, just more artificially looking. For comparison again the same detail as it came out with the program mentioned before, Neat Image - again without and with grain removal algorithms applied
What happens, if one removes all grain and grain like structures - the image becomes unreal and artificial - here the result on the right side with some more detailed optimization on the left side below
Finally a last experiment - how does this same detail look without that much grain and shot with a film which has not such a high speed? I have shot that picture below with an Agfa Optima 200 film ( ISO 200) and taken the more or less same detail of that red onion (in the meantime, the onion started to dislike being cut open as you can see it on the increasing gap between its layers) - my conclusion is clear: Take the lowest possible film speed and you will get much better color rendition, much less grain, images which are better in resolution and overall more pleasant. But if you need to shoot with a high speed film, then you got some tools to make your images appear more pleasant.
SILVERFAST AACO What does AACO stand for? It is a tool which does a contrast optimization. Very often, photographers face the problem that the sun is shining out of a completely clear sky, no clouds are reducing the overall contrast and therefore the film (and of course also the digital sensor) will not be able to cope with the contrast range of the scene. There are a few options to cope with this situation like returning another time when the sun is partially covered (this would be fine but time consuming and often not possible), taking two or more shots with different exposure time and later compose one single image out of two or three, taking always the best part of one image and pasting it into a new one, or, as a compromise, living with what you have got and optimize the result by applying a contrast expansion routine during scanning. It is clear, that this technique works only if there is something to be scanned - a clear part in a negative or a black one in a slide will never produce anything, but if there are details still visible, then AACO can do some magic. I tried it out and it works fine if used properly. How does it work - it works (for the user) similar to the shadow-highlight function in Photoshop. The shadows are brightened up - the amount, the contrast range, the color range etc are user definable. Let me show you an example: The first image, a shot from the Grand Canyon, was taken on Ektachrome 100 on a very sunny day with the sun shining in an unfortunate angle for the position but creating attractive highlight-shadow contrasts.
As you can see, the darkest parts in the foreground are much brighter but still look natural - I did not apply too much of this function as it can create a very artificially looking image if the amount of enhancement is overdone.
Another example comes from a negative scanned on an Epson 3200 Photo. The location was one of the most frequently photographed places in Colorado, the Maroon Bells. In June, sunrise happens to be such that the rising sun shines exactly into the middle of the valley, which means that the Maroons are evenly illuminated. The AACO function needs some fine tuning to become a real benefit - the two parts of the scan below show the effect of AACO - left with AACO activated and manually adjusted and right the same part without AACO. The third scan is showing what happens if you take AACO in automatic setting and do not optimize its settings for the problematic zones. I admit, this negative is a pretty difficult one to scan and to optimize as the contrast difference between the Maroons in full morning sun and the rest of the valley still in full shade is tremendous - but with AACO you get a nice and acceptable result.
This function is pretty good, works on all my scanners but is only available for the "Studio" version as I was told. So, you need to get this more powerful version to benefit from it or as alternative, you have to use Photoshop and its built-in Highlights-Shadows function ( which also needs manual adjustment to work properly ),
SILVERFAST MULTISAMPLING The last function which I would like to review is the multisampling function. Multisampling means that the scanner scans the selected area more than once. This feature is especially for lower end scanners quite interesting as it increases the overall quality of the scan and reduces scanner inherent and image related noise. Scanner inherent noise is often visible at lower end scanners with simpler electronics and scanning head and of course in those image areas where the details are not very well visible like in dark areas. But as we shall see, this multisampling is also not without some problematic side effects. Let me start with a color negative which I shot several years ago in Australia - a Koala sleeping and sitting high up in a tree in a rather darkish area. I had enough troubles to hold my Mamiya RZ67 steady enough to get a sharp in-focus image and hoped that the mirror upswing would not introduce vibrations - well it went fine but I had to take the shortest possible exposure time which created a slightly underexposed negative ( Kodak VPS ). The overall image (in which I marked a square area red) looks like that:
I scanned that negative twice - once the standard way, so just 1x and then again without any changes 8x - and I hoped to see the difference in that mentioned square area. After enlarging that small area and putting the two scans into one image for better comparison it looks like that:
As you can clearly see, even here on the web, in the center, where there are no more details in the negative, both scans show just black "noise" but outside, where there are some details still in the negative, the 8x scan reveals them better with a less grainy substructure - best visible in the lower part in the structure of the tree. But - and this is visible as well, the overall definition of the image suffered a bit - it appears to be less sharp. Why - well let's think for a second about how the multisampling is achieved: The scanning device moves from point zero 8 times over the selected area which means that if the scanning head does not start 100% exactly at the same place of the image, then the 8 scans will be slightly different as their starting and ending point is different - maybe just a fraction of a ten's of a millimeter but in this large magnification, it is visible of course. What one can do is to apply un-sharp masking to increase the perceived sharpness. This works to a certain extent but also this masking technique has got its limitations as it increases again the grainy structure as well. Let me show you another example. The photograph below was taken in an ancient lava flow in central Australia. The location was lit somehow with a couple of Halogen lamps and I was able to use a tripod. Due to the large difference between the light inside the cave and outside, the outside part is overexposed but to get at least some details, I had to underexpose the inside quite a bit. This resulted in noisy scans in many parts of the negative (again Kodak VPS). Here is the photograph itself:
To improve the overall quality of the scan, I have applied 8x and 16x multisampling. The difference between 8x and 16x was visible but not extremely significant. Another difference however was much more visible - the comparison of an outstanding dedicated film scanner ( Nikon Super Coolscan ED8000) and a very good flatbed scanner (Epson 3200). The following three details show it clearly - and here you can gain much more by using a dedicated excellent film scanner than using a multisampling option. Of course, if you have got a flatbed scanner which is fine but not outstanding for film, then multisampling is an interesting and attractive option. The two images show a small part of the background, the left one multisampled 8 times and the right one 16 times.
But if you scan the same negative with the ED8000, the result, even for a single scan (1x sampling) is outperforms by far the multisampling results obtained with the Epson 3200. Here is the same area scanned with the ED8000 - no corrections made, no adjustments, no contrast optimization, just a 1x scan:
My conclusion is, that if you have got a simpler scanner, then multisampling can improve your results significantly and it is very worth applying - despite the time needed for that procedure - but as expected, if you use a very good film scanner, its quality is just better. Of course, you could now conclude - how must be multisampling with such an excellent scanner. Well, again, as expected and shown before (the Koala was scanned with the ED8000) - as long as the negative shows some details, multisampling can get them into the scan, but if there is nothing, then even the best scanner cannot improve the results. CONCLUSION SilverFast 6 is a rather expensive but very powerful scanning software. When I got it first, I did not like it very much as you can do many things wrong and the hope of just pushing a button to get perfect results with an all-automatic setting is not realistic. You need to understand the in's and out's of what influences the results, what do certain parameters mean, how do they interact and where are their limits to get, after quite some trial and error, to good results. This is something which a lot of people overlook and as a result they are disappointed and frustrated and blame either the scanner or the film or both for their very average results. What to say now - SilverFast or not? Well, it depends on what you want to achieve - do you want to quickly convert a negative, a slide or a print into something digital or do you want to get the best possible conversion into a digital file from a given target? As I like cameras very much, let me apply a comparison using cameras: You can shoot a nice picture on any simple camera - several years ago I would have said - like a Kodak Instamatic - but if you want to capture a scene with as much details and accuracy as possible, you should aim for a real camera. There are so many features, options and other goodies in SilverFast that it would be too much to describe all of them - what I wanted was to review a few important ones which have a significant impact on the outcome - a perfect scan.
|
|
Send mail to
Webmaster with
questions or comments about this web site.
|
