Auto/manual focus with Minolta Elite Scan 5400

Hi,

 

does anybody know how the (auto)focus of the 5400 works? Does the

scanner look for contrast in the (pre)selected area of the picutre

(focussing area)or how does te scanning mechanism detects sharpest

focus position?

I'm using Minolta's scanner sowtware, vers. 1.1.3. Sometimes I get

sharper output with choosing focus area at a different position out

of the center of the picture. Nevertheless I never managed to get

the bar in the manual focus display in a position right to the

middle.

The velvia slides are extemely sharp (tripod, hight quality lenses

(100mm macro, 2,8 200 APO G) and I think the scanner cannot

reproduce it. How big is the focussing area really, what are you

experiences in finding the optimal part of the picture to set

focussing area?

 

Please try to give some advice concerning the Minolta software.

 

Thanks in advance!

 

E.G.

I have the Scan Dual II. I do sometimes notice a sharper picture by manually focusing on an object with good detail, perhaps even what I focused the camera on originally. As I recall, best focus is achieved by the moving the focus bar as high as possible. This usually requires a little jockeying back and forth, much like focusing a camera.

I suggest manual focusing outside the middle of the image on

an image part that does NOT show any sharp details. This way

you force the focusing system to look at grain contrast and it

cannot be distracted by actual image info. Look for mid-grey even

areas, possibly unsharp in the photo, but of course showing

sharp grain.

@Brian

Your method is exactly what I usually try. Sometimes it really helps.

 

@Erik

The point you bring into the discussion is a very very interesting aspect for me. I never took this method into account before and will try it the next days.

Some additional questions to make sure that I will get the expected better results. Will I have to scan at 5400dpi to make the scanner "see" the grain and will I have to disable grain reduction? Usually I use grain reduction and scan at 2700 dpi.

 

E.G.

"Please try to give some advice concerning the Minolta software"

 

Drop it and try Vuescan.

"Will I have to scan at 5400dpi to make the scanner "see" the

grain and will I have to disable grain reduction? Usually I use

grain reduction and scan at 2700 dpi."

 

E., You can leave the grain reduction on, although this might

make it more difficult to focus. Minolta's so called Grain

Dissolver is inspired by my Scanhancer for the Minolta Multi Pro

(see my website for further info). It is a hardware diffuser to

de-collimate the light that goes through the film. This is better

than any software grain reduction (which you could still apply if

you wanted, afterwards). I would however scan at 5400 dpi (16

bit) and if needed downsample in Photoshop. Downsampling is

what Minolta's SW is doing too, but it does a less good job than

Photoshop, beause it makes the scan head move with bigger

steps in one direction and downsamples the resolution of the

line-CCD in the other direction. Because of this scanning

method actually less input information is used to calculate the

downsampled information from.

Some comments:<br>

"Maximizing the focusing bar" during manual-focus is typically the kind of action that can be automated very easily.<br>

=> I am not quite sure that you can get better results by doing it manually, rather than with AF. I would be interested by more data/experience.<p>

 

Erik: The Elite 5400 has a very specific way to handle scan and downsampling:<br>

- a "5400dpi" scan will use the 5300 cells of the CCD line(x3 RGB) and make 7800 steps to move the line (="full use").<br>

- a "2700dpi" scan will also use the 5300 cells (x3RGB) and make 7800 steps, thus handling the same amount of data ('scan time'+'CPU usage' and tests support this, as you can read in this <a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=006Heo">thread</a>). So the scanner is actually making a 2x2 downsampling, just like Photoshop - maybe not a lanczos, though...<br>

- a "1350dpi" scan will use 2650 cells and make 3900 steps (the scanner itself might gather data from 2 adjacent cells, meaning in reality 5300 cells being used, but I seriously doubt it).<br>

- There is always this 2x2 downsampling, except with ICE+GD and smallest resolution (675dpi) where you get an 4x4 downsampling.<br>

=> Bottom line: there is no need to scan at higher resolution and downsample latter, as DimageScan is already doing it.<p>

 

You're advice to focus manually in area with low contrast from the image, in order to pick the contrast of the grain itself is very interesting. I will give this method a try. Thank you.<p>

 

Olivier

Advice: make a Scan at 5400dpi with Auto-Focus on a part of the image with a strong contrast (AF needs it). If you do see clearly the grain <b>at this very position</b> on a 100% close-up, then it means that the AF is spot-on and sharp.<br>

Now, if you don't see the grain in other parts of the image (corners, edges, center...), then you are dealing with a Depth Of Field issue for the scanner, and not with a focus issue.<br>

When I AF in the center of the image, I can get soft edges because of this DOF issue, even on moderately curved negatives... It is better to set the Focus spot closer to an edge, in order to optimize the use of DOF.<p>

 

Olivier<br>

PS: Mendel, I had a laugh with this excellent and detailled comment from you :-)

Olivier,

this discussion produces some good information for me. We are talking about areas with high or low contrast. Are we talking about very small areas with only a few pixels, e.g. an area with high contrast would be one with a single very thin (1mm or less on monitor) black line onto white background? I don't understand how big the area is that the scanner analyses for autofocus.

 

We have three possibilities now for further testing:

1. the scanner doesn't use contrast of the picture at all, just grain is important for autofocus. the part of the picture we have to select for autofocus should not be too dark.

2. grain is not important. we have to select parts with high contrast.

3. grain is not important. we have to select parts with low contrast.

 

the 3rd possibility doesn't make sense in my opinion.

 

E.G.

I believe that the scanner uses contrast in order to focus.<br>

An easy/good situation for such a system is a sharp separation between a black area and a white area (you should center the focus area on this separation), and there is no issue on the size of the focus area. Lines and strong details are also a good situation for such a system.<p>

 

The focusing system uses the contrast on the specified focus area (I don't know the size of it), that comes from the film (image and grain pattern included). It is closer to your point [2].<p>

 

However, the experience from Erik can make sense, as it forces the focusing system to consider ONLY grain: the contrast is low, but it is directly related to the smallest element on film => focusing is difficult, but if you can really maximize it, you are quite sure to have excellent focus.<br>

Some (negative ;-) thoughts, though:<br>

1. the focus system must have a limited accuracy, and by lowering the contrast level (grain), you may end completely out of focus.<br>

2. if the system is able to handle properly this low contrast level, it should also be able to handle such a small variation of contrast for higher levels of contrast + the same improvement on focus will give more difference on contrast for a high contrast situation than from a low contrast situation => better results on high contrast part...<br>

This being said, Erik has an EXCELLENT record on scanning issues, and I quite believe that he has done his homework here as well => So I will spend some time testing the focus system of the Minolta 5400, and report my results later...<p>

 

Olivier

Thanks for the compliment, Olivier! It is my experience with the

Multi Pro that manual focusing on grain-only works most reliable.

Focusing on the grain is only possible with scanners that clearly

resolve grain, but I consider the Elite 5400 to be such a scanner.

When focusing at contrasty details I noticed that it was

sometimes very difficult to find one point of maximisation of the

focus bars. When focusing on blurry areas maximum focus is

always easy to find and also very accurate, to my experience.

Autofocus has never given me such accurate focus, probably

because it takes bigger steps during focusing then when slowly

operating it manually. Unfortunately the Elite 5400 has no glass

holder, so the curvature of the film might change under the

influence of heating up during focusing or multi-sampling.

 

In case you want to make your own glass holder: use the special

high pressure anti-glare coated glass used for expensive gallery

frames. Do NOT use glass with anti-glare etching! The anti-glare

coating also works as an anti-Newton's rings layer, but does not

degrade the image quality, unlike etched glass. Only a slight

colour shift may be expected because a little more UV will be

blocked.

Erik, some very deep information again. Thanks again. I will report my test results.

A comparing test in a (German)magazine about Elite 5400 and the new Nicons V and 5000 pointed out that the Minolta has a larger DOF than the Nikon Scanners have. The interpretation according to the test was that the 5400 has minor problems with curved slides compared to Nikon. They stated that Nikon is a bit sharper with flat slides but has some trouble with curved slides. Maybe the 5400 has a different DOF compared to zthe Minolta Pro scanner, too?

 

However, I heared that high quality scanners don't produce absolutely sharp images because this empowers the scanner to get more information out of the slide. Maybe there is a relation betweenn higher DOF and lower maximum sharpness level (but higher sharpness in average with curved slides).

 

E.G.

I made several scans to check focus accuracy (at 5400dpi, on 3 different color negatives). Here are the results:<br>

- on 1 photo, without GD, I had MUCH better results with manual-focus than auto-focus (check 100% crop at the focusing point). I tried on 2 different spots in the same area => same issue with AF accuracy.<br>

- for other photos, I got exactly the same focusing and results for MF and AF.<br>

- The software system (arrows) was much more accurate and convenient than the dial for manual focusing.<br>

- The contrast indicator (white=max/black=current) was longer, thus probably more accurate, when used on a clear area (cf Erik's recommendation) than on a contrasty one.<p>

 

Because I had 1 photo where AF did failed repeatedly (and quite noticeable effect), I have decided to stick to Manual-Focusing (with spot = 1/3 from Corner to Center of the photo).<p>

 

Other persons have reported having problems with AF for some photos...<p>

 

Olivier<div></div>

I changed to MF completely. The reason for this is that I noticed that focusing works best if I set the focus point to areas of homogenous colours. It seems that the 5400 analyzes the grain of the film in a realtively small area and succes of focusing depends on the colour, too.

In the beginning I tried to set the point to areas that apear sharp in the images, e.g. the border of a window. But this is useless because only the grain is important.

 

I get the best results if I set he point to areas with medium grey or brighter colours, e.g. a bright blue sky. The red coulour seems to cause some problems. I couldn't analyze that the position in the picture, e.g. the center or somewhere in the middle of the photo for itself influences focusing for at all.

 

E.G.

Olivier and E.G.,

 

I am happy to read that we agree on the correct method of

focusing with a Minolta scanner! Hope this matter is settled now,

as I keep reading questions about it in many other places.

I have tried different methods on a dozen test slides (mostly Fujichrome) to get the best focus on a 5400, using the following subjective and unscientific process.

 

- Slides are mounted on Wess or Gepe mounts which flatten the film much better than the conventional mounts.

 

- On each slide, several raw scans are captured at 5400dpi with different permutations of manual (in sw and NOT with the knob) vs auto, center vs off-center, high contrast vs low contrast, first frame position vs other positions.

 

- The scans' sharpness is evaluated by eyeball side by side in PS at 100%.

 

My observations:

 

- Auto focus works better than manual focus on some slides, and vice versa.

 

- Focusing at the center works better than focusing off center on some slides, and vice versa.

 

- Focusing on a high contrast area (tone or color) works better than focusing on a low contrast area (Eric's suggestion) on some slides, and vice versa.

 

- A good focusing method (a permutation of the above) on a slide in the first frame of the carrier will not necessarily provide the same result when the same method is applied to the same slide moved to a different frame position on the carrier.

 

My conclusions and suspicions:

 

- There are sufficient variables involved that I cannot arrive at one single method which can produce the "best" focus on every slide.

 

- The size of an area being focused (auto or manual) is unknown, but I suspect that it is rather small.

 

- The Minolta (or other third party sw) window is not a good place to evaluate the focus sharpness.

 

- The manual focus slider bar extension varies from spot to spot on the same image. It seems to extend farther to the right on spots with a higher contrast. I always use the one that extends the bar farthest to the right.

 

- The positioning of the carrier is uneven from frame to frame.

 

- The 5400 has a shallow DOF, but it is unknow whether it is better or worse than the Nikons. There are definitely more written complaints from Nikon users on this problem (due to larger user base?). Some Nikon users attribute the problem to the LED light source.

 

- A scan can be sharpened by the scanner sw. A focus evaluation must avoid this variable.

 

For critical scans that I will apply significant editing efforts in PS for prints up to 13x19, I bracket focus my scans with the following workflow.

 

- Capture a few raw scans with a few permutations of the above methods. One of the methods will focus on an area of an image that has/needs the maximum sharpness, e.g. eye lashes.

 

- Evaluate the scans in PS at 100%, and select the best one to work with.

 

Getting a correctly focused scan is very much part of "getting the best scan possible" adviced by many books and tutorials. But I have yet to find one source that touches on it in some detail. It is also a key feature of a scanner. Yet the manufacturers only want to boost their resolution and Dmax, and don't seem to have good solutions for a better DOF and better carriers. What is the point of a scan with high resolution and wide Dmax, but is not critically focussed sharp? Hope they are reading this.

Well, some months since the last word but... I am using a Minolta IV scanner and found everything said above to be true. I am talking more about negative strips than mounted slides (as I use much more B&W film).

 

PS: I know the discussion is about Minolta 5400 but maybe it helps too.

<blockquote>

<p>I get the best results if I set he point to areas with medium grey or brighter colours, e.g. a bright blue sky. The red coulour seems to cause some problems.</p>

</blockquote>

<p>Interesting! I can also confirm that focusing on red gives bad focus (using the manual focus meter)! But this is on negative film.</p>

<p>E.G. ... were you talking of red in slides or negatives?</p>

<p>Is anyone even reading this 5 years after the fact? :)</p>