samagnew

<p>Just as an aside, I assume the difference with the medical scanner is that it is for X-Rays and therefore presumably monochromatic. How awesome would that be? You should get much better scans of black and white. If you wanted colours I guess you could just do three different scans with three primary colored gels over the backlight unit and combine in Photoshop. More tedious, I guess, but imagine the fine control you could achieve. You could probably get some really cool colour effects just by using slightly different gels for the different primaries.</p>

<p>Very intriguing... Very off-topic but very intriguing!</p>

<p>Sam</p>

<p>Wolf,</p>

<p>I think we all understand your frustrations if you made such a bold decision as to sell your 9000 in advance. It must be very aggravating. However, I think most of us would agree that this would be classed as "taking a chance". Lots of people are following all of this buildup and many think we would be interested if it really does match or outperform the big Nikon. However, I think most of us are likely to wait on a shipping product and real test results before pulling the trigger. That you didn't wait is certainly bold but I think you should be big enough to live with its consequences. Mark and Plustek's launch problems are their own and your self-created issue is your own.</p>

<p>Just wait and see like the rest of us. Maybe it will be great. Maybe not. I think it could be great. But I'm not letting go of my 9000 just yet.</p>

<p>Sam</p>

<p>In my opinion the world was not short of "OK" photo scanners. Look no further than the Epson line of flatbeds with tranny adapters. What the world lacked after the passing of the Coolscan 9000ED was a properly "enthusiast" GOOD film scanner for under $10,000. I sincerely hope this OpticFilm 120 is that. If it is, it will be more than worth the $2000.</p>

<p>Sam</p>

<p>I guess what I really hope is that between the current renaissance in "artistic" photography and the curiosity of a younger generation I'm hoping that a truly good new scanner would have the power to re-invigorate film shooting. I certainly am someone who is scanning their current output, not their historic output. I shoot film not because I don't have the choice of digital but because I choose film.</p>

<p>However, I want it to look good in its digital form since that is how images are shared and even printed these days.</p>

<p>Is there a problem? I find it curious to still hear nothing from Plustek.</p>

<p>Sam</p>

A drum scanner won't have a warranty or ICE and will require you to maintain an obsolete computer to run it. If your

budget is $2000 I would strongly caution you against a drum scanner. I would love to have one also but Tim Parkin is the

next best thing to owning one (cheapdrumscanning.com).

As for the Plustek, I will be more excited if it is closer to $2000 providing that the quality and durability implied in that price

are there. The world of film really does need a replacement for the late great Coolscan 9000.

Wow, is that --- seven holders? Are Plustek really looking to pick up the Coolscan 9000's crown? Fantastic! I hope this beast is as good

as it looks!

<p>Mark, when will we see actual specs? When will there be units available for review? When will they go on sale at major resellers like B&H?<br>

<br />Sam</p>

<p>Let's be realistic. If it is going to be any match for the Coolscan 9000 it is going to cost at least as much. The 9000 went for $2200 new as I recall. Also, it will be plastic but from the product shot we saw it is a very good looking box. The trays will be plastic. They were on the 9000 and every other scanner I recall. If we are lucky they will be well designed to hold the film flat and there will be a selection of different ones available. If it costs $1000 it won't be touching the Coolscan for performance. How could it really be otherwise?</p>

<p>I for one just hope it will be great. If it is great it will be worth somehow affording. If it is too "affordable" then it will hardly be worth buying.</p>

<p>One feature I think is really strong on the 9000 that I don't see on the other Plustek and cheaper scanners is an automatic transport. I think this is key to productivity. The 9000 is not the fastest in the world at 35mm but at least I can load twelve frames (strip film) into it and not look at it again until I need to load the next twelve. The same with strips of four frames of 6x4.5 120 film. This batch capability is key (for me) to getting through a stack of film. If I had to advance each frame by hand I would be a lot less tempted by the product even if the IQ was great.</p>

<p>My $0.02</p>

<p>Sam</p>

<p>Rishi,</p>

<p>Yes, I think you understand me perfectly. What I am saying is that I see very smooth tonality in scans made with better equipment (drum scanners) and that oversampling helps on something like the 9000. And this makes me think that the sensor/A/D chain is still the limiting factor.</p>

<p>I also LOVE slide film. I think negative film is important for the reasons I gave before and also because of the ease of home processing or mini-lab processing. It is much more common and accessible so the market will judge a scanner largely based upon its ability to do negatives well. Or, conversely, if it is awesome at negatives that will be what will make it a legend alongside such greats as the 9000.</p>

<p>Sam</p>

<blockquote>

<p>It's not an impossible task. It's just that no one's had the resolve <strong>AND</strong> resources ($$) to do it yet (albeit: it'll be easier when we have CCDs with 14-15 stops of dynamic range to deal with slide film which, IMHO, is the only film anyone should be shooting today b/c the S:N ratio of negative film is just unacceptable next to digital, despite its magnificent dynamic range).<br>

-Rishi</p>

</blockquote>

<p>Actually, I feel that this is film's sweet spot. Hence the earlier discussion.</p>

<p>Negative film does trump digital in two of its very weak points. Dynamic range, and highlight handling. Of course the tradeoff is that this means that the dynamic range is very compressed on the film itself. A poor quality scanner will turn this into poor signal to noise ratio and a very visible lack of "smoothness" in ranges of even tone.</p>

<p>This is why a good sensor and good signal path and good A/D converters are critical to getting good results from it. However, on a scanner that can handle it (the 9000 is just about good enough, drum scanners are great at this) negative film shows the strengths of film and a hybrid workflow like no other medium. Hence the passionate discussion about the need for consistent tools for properly inverting the negative image whilst preserving the image's integrity.</p>

<p>Sam</p>

<p>This is a heartening development. Glad I'm not just speaking in a vacuum here.</p>

<p>I agree with the above sentiments. There is absolutely NO reason that negative film cannot be as colour consistent from frame to frame as slide film. It is still high quality dyes on film and x mix of light will produce x mix of primary colors. The problem is in the history of the products. Slide film was used for publication and has a whole history of being painstakingly colour matched on output for WYSIWIG. Color neg was always hugely popular with consumers and the traditional reproduction involved colour balancing by eye for a nice final image rather than for accuracy (e.g. every corner photo lab).</p>

<p>It is an absolute waste of time for each of us here to laboriously re-invent the wheel from first principals. It is shocking the number of charlatans who make their money by claiming to have solved the issue when they are merely peddling another "close enough" solution. Silverfast, negfix8, ColorPerfect, etc. etc., I'm looking at you! The people who come closest to getting this right are Vuescan who at least have an option to make getting a good usable raw scan easy (the old "lock film base", "lock image colour" two step. If you select to also have a "RAW" output it will give you a non-inverted TIFF that should have all three channels with the same exposure on each frame).</p>

<p>ColorPerfect. They are so close! I've gotten some brilliant results in the past but they are another purveyor of auto-white balance magic. You can sometimes get great results but never consistent ones. This is because they seem to want you to edit to a final image within their plugin rather than allowing you to get an accurate "flat" conversion. </p>

<p>My current workflow involves finding an appropriate scanner exposure for all three channels and fixing it. One that gives me no clipping on either end of the scale for all channels.</p>

<p>Next, I scan all frames in the roll at this same fixed exposure and try to take at least one picture of a neutral greyscale target on each roll.</p>

<p>Next I bring in the scan of the "end" and go through a set of actions called "CNeg" from an old free set of photoshop actions called "DonzRGB". This simply automates creating a bunch of layers with levels and curves controls with helpful labels.</p>

<p>Next I set "black" and "white" from the "end" scan frame and leave all the other layers alone.</p>

<p>Next I copy this layer set to the frame with the greyscale and do whatever I need to with the "curves" layer to get a neutral grey scale.</p>

<p>Finally, I take the layer set from this frame and copy it to all the other frames. This gives me consistent accurate colour on all the frames in that roll and a good base image for each frame that I can then make artistic adjustments to.</p>

<p>But what a lot of work! Both to come up with the workflow from first principals and then to manually implement. Someone could automate this if they cared. If they did then the vast bulk of "new wave" film shooters might find themselves with excellent images rather than mostly-disgusting-looking "retro" images. This would raise the puplic perception of modern negative film from a curiosity to a real contender.</p>

<p>Someone should automate all of this as part of the scan process. Photography and the future place of film in the public imagination deserves it. The company that makes it easy as part of their scanner solution could make a whole new market. Rather than merely add a little dpi here and a little dmax there they could have a product that fundamentally improves most people's scans!</p>

<p>Lest anyone think this is all theory and hot air, these are the recent images I have created using this workflow. What you won't see, unfortunately, is the greatly reduced effort overall in getting to a good pleasing rendition versus the many other avenues I have been down in the past:</p>

<p>

<p>Also just about anything recent in my stream (2012 or newer). I want to specifically point out the last image of the studio shot of the cupcake with the white background. I challenge you to take a shot like that through the scanner software of your choice as "negative", through negfix8, through ColorPerfect or anything else and come out with something that isn't mush! It would be fantastic if this could be a properly solved point-and-click solution for all of us. Kodak, after all, are making some AMAZING colour negative film these days. Aparently, they have even now found out that it is the business they are actually best at.</p>

<p>As I understand it, the exposure for each R G B colour in a scanner is usually varied simply by varying the "dwell time". So the aperture is fixed, the sensor gain is fixed, but the time the film is exposed to each colour sensor varies. To double the exposure for blue, for example, the scanner merely exposes the film to the "blue sensor" for twice as long as to the "red sensor". This is a very simple procedure and as far as I know is common to all scanners that can scan negative film.</p>

<p>For example, if you scan in "negative mode" the scanner normally works this all out for you. However, if you scan in "positive mode" you normally have to set the three different exposure times or "analog gains" yourself.</p>

<p>So it's not the hardware that needs to change, it's the scanner software. And I fear that the market is too small these days for the likes of Plustek to invest in creating their own scanner software. They will probably bundle Silverfast or something like that.</p>

<p>Sam</p>

<p>I would do the exposure for R G B based on mask in scanner software as you are actually setting your scanner exposure to capture maximum information. I usually leave a little margin to avoid clipping black (positive black). Apart from locking exposure I wouldn't do anything else in scanner software. Curves and white (positive white) I would set as layers in Photoshop and apply them to all frames after scanning.</p>

<p>I feel a little bad about the thread highjacking. I was really just quite moved by Paul's images and felt moved to respond regarding his frustrations with inverting consistently. I'm really excited about this scanner. If it is good it will be a fantastic resource and I really hope is sells well. If it does, I wonder if Plustek would punt on a real film scanner like this for 4x5. It has been a very long time since the last film (not flatbed) scanner was available that would do 4x5. It is only a little bit wider, really, than 120 film...</p>

<p>Sam</p>

<blockquote>

<p>How would my endpoints differ from the 'real' endpoints if I'm scanning a <strong>blank</strong> frame of negative film to determine the true black point?</p>

</blockquote>

<p>Re-read what I wrote. The endpoints on the "end strip" (a frame with some fully exposed and some unexposed film) WILL be the true endpoints. The endpoints on any random image from the middle of the roll will not. If some unexposed film is present at the margin then you may have the opportunity to sample that as a "true" black but you will still be guessing at the white based on whatever is brightest on that particular frame.</p>

<p>This is the problem I am discussing.</p>

<p>The manual workflows described are pretty much a manual way of doing what the automatic software does. The key step is where you take the three colour channels and adjust the endpoints according to the histogram. This step is where you (like the automatic software utilities) are making an assumption that the brightest bit for each channel is essentially white and the darkest bit is essentially black. If you have shadows and/or highlights which are less than perfectly neutral you will be introducing a colour cast that will be troublesome to remove. Since your endpoints now differ from the "real" endpoints of the film (in other words, you have perhaps unwittingly defined very dark brown as "black" and very bright pink or yellow as "white" and remapped the image to these black and white points you selected when you defined the endpoints for each channel) the correction required to correct any cast is no longer a simple gamma curve.</p>

<p>What is far preferable is a stored set of endpoints determined from a strip end as well as a stored set of gamma curves determined from applying those endpoints to a greyscale image on the roll and colour balancing it with per-channel gamma curves. The endpoints plus curves applied to any other frame should yield accurate colour.</p>

<p>To put this another way. Take a slide scan or digital camera image that is already positive and colour accurate. Note the per-channel histograms. The endpoints for each colour are not necessarily the same. If you make them the same (as per the workflows referenced or as the various softwares discussed to automatically) you will be degrading the colour integrity of the image as you will see if you perform the experiment.</p>

<p>As far as the character of different films. They are not, in my experience, as easy to remove as that. If that were the case, every colour-accurate slide film would look like every other one. Bad colour accuracy is not the defining character of any decent negative film.</p>

<p>Paul,</p>

<p>I have spent quite a large amount of time on this question of inverting negatives. The short answer is that you can't get consistency because all the software vendors are trying to give you an automatic solution.</p>

<p>To get a consistent conversion you need to know the inherent balance between the three colour levels in the film (traditionally you could get this by calibrating the scanner to an unexposed portion of the film). This then also gives you a true "black" (negative film is inverted). For a true white you can come close by sampling a fully exposed "end" piece of the film. Whilst it is possible that this represents a true white there is a possibility that one or more channels may suffer "solarization" and that this may be slightly off. In my experience with modern films this isn't something you really need to worry about, you can pretty much take this as "white". Lastly, you need a grey scale image (you can photograph a target with daylight balanced light source) to figure out the curves. You can then have the "black point", "white point" and "response curves" as layers and apply them to all your frames. You should have consistency from this workflow under one condition. The condition is that the scanner settings do not change from frame to frame. Since the scanner will always do its own adjustments if "negative" mode is used you must manually scan in "positive" mode. Not only that but you must fix the exposure since most scanning software will do an "autoexposure" for each frame. Lastly, your chosen manual exposure must be sufficient to correctly scan all the frames without any clipping of highlights or shadows on any of the frames. This is a little harder on a device with a narrower DMax range.</p>

<p>What negafix is doing, just like Epson Scan, like Nikon Scan, like Vuescan, like ColorPerfect, like everyone is that they are taking an image that generally has no unexposed or fully exposed section of film showing. This means there is no hard reference since no film border or fully exposed film is in the frame. Even if some film border is included in the image the software is generally set to ignore the edges of the image so as not to be confused by the image of the film holder itself. Clever software is then guessing at the endpoints of each primary colour based on the idea that the brightest highlights and darkest shadows are generally neutral. It is then applying canned curves for whatever film brand you told it. Since the endpoints are based on guesses based on what is actually in the image area of each frame the results change from frame to frame.</p>

<p>I understand the motivation of the various vendors in trying to simplify things for us but I would dearly love someone to automate all of this at the scanning stage so that we all could get truly consistent colour from our negatives.</p>

<p>By the way, I love your results. I would also love this scanner to be as good as or better than my 9000. And I would REALLY love a solution that would automate what I described above but, sadly, it would probably be a hard sell since the results of the "automatic" products seem to satisfy most people.</p>