Thinking of Nikon ES-1 or ES-2 for negative conversions- Do I need a light box ?

[ross_lipman]

I have a D7100, and several macro lenses. I may purchase a 40mm as recommended.

 

What should I use for a constant light source? Do I need a light box, or can I use something more readily available ?

 

Adding the cost of a good light box increases total cost to close to current value of a used negative scanner.

 

What are others using ?

 

Thanks.

[Ed_Ingold]

You do not need a special light source to copy slides or (ES-2) film. Any reliable light source will do, including a flash unit. I use a desk lamp with a "daylight" LED replacement bulb, bounced off a white card for additional diffusion. Since the ES-1/2 is part of a rigid assembly to the lens, a long shutter speed can be used. With the desk lamp, my typical exposure is 1/4 second at f/5.6, ISO 400.

[ross_lipman]

Thanks.

 

I am surprised you use 5.6 and 400. Why not f16, 100 and a longer shutter speed ?

 

I can easily set up a white light and diffuser.

[Ed_Ingold]

There is very little difference in quality between ISO 100 and 400. F/5.6 gives the best resolution from the lens, a 55/2.8 Micro-Nikkor. The camera I use is a Sony A7Rii (or A7Riii), which allows very fine focusing through digital magnification - live view on steroids. I left the borders of the slide mount for illustration purposes. Exposure and color balance aren't affected as much as with color negatives.

 

The magnification is slightly less than 1:1, in order to allow fine focusing with the lens barrel.

 

Ektachrome Copy with Sony A7Rii + Nikon 55/2.8, f/5.6 @ 1/4 second, ISO 400

Tom Sawyer Day in the Country, Indiana 1987

[rodeo_joe1]

Do I need a light box, or can I use something more readily available ?

No, a lightbox is definitely not needed. Both the ES-1 and ES-2 have built-in light diffusers. All that's needed is a light source of approximate daylight colour temperature.

 

I use a camera-mounted flash bounced back off a white board set about 1m (3 foot) in front of the camera. 1/4 or 1/8th power on the flash is all that's usually needed. The only 'drawback' is the lack of light for focussing, so you need some continuous source for that.

 

Other people successfully use LED lamps, which incidentally flicker at double the mains frequency. LED lighting obviously needs a longer shutter speed to cover the flicker frequency, and because of the lower intensity.

 

Here's an example of the results I get from scanning colour negative film using a device similar to an ES-1/2:

Edited May 12, 2019 by rodeo_joe|1

[rodeo_joe1]

PS: If you do get an ES-1 or 2, make sure the plastic diffuser is totally clean on both sides, otherwise you'll get a dark spot on your slide scans - or a light spot on your neg scans.

[Ed_Ingold]

Other people successfully use LED lamps, which incidentally flicker at double the mains frequency. LED lighting obviously needs a longer shutter speed to cover the flicker frequency, and because of the lower intensity.

LED replacement bulbs, which resemble incandescent bulbs, operate by fluorescence, excited by a near-UV source. The retention time is almost a second, which is more than enough to eliminate any flicker.

[rodeo_joe1]

The retention time is almost a second, which is more than enough to eliminate any flicker.

No, it's not. If you point a camera with focal-plane shutter at a mains driven LED lamp, you can clearly see an exposure variation across the frame at high shutter speeds, and a variation of exposure between frames at mid speeds.

 

With almost any mains driven lamp, it's advisable to stick to shutter speeds that allow at least one half-cycle of mains frequency to pass during the exposure - i.e. 1/100th second where the mains frequency is 50Hz, and 1/125th where it's 60Hz. Preferably a shutter speed that allows one whole mains cycle to pass, or where the shutter blinds completely clear the frame - X synch speed - because colour temperature usually changes throughout the cycle too.

 

Fluorescent excitation dies away exponentially, and although the afterglow may remain visible or detectable for some milliseconds afterwards, its intensity dies very quickly with removal of the primary source.

 

To the OP: I see you're using a DX camera, which will require extra distance between the front of the lens and the slide. The ES-2 was designed specifically for the full-frame D850 and may not have sufficient adjustment to use with a DX camera. Check before you buy.

Edited May 13, 2019 by rodeo_joe|1

[Ed_Ingold]

With almost any mains driven lamp, it's advisable to stick to shutter speeds that allow at least one half-cycle of mains frequency to pass during the exposure - i.e. 1/100th second where the mains frequency is 50Hz, and 1/125th where it's 60Hz.

My exposures are on the order of 1/4 second, or 15 cycles (US). I never experiencedany banding nor variation in exposure across the frame. Nearly every venue I shoot for video uses LED illumination, always at 1/60 second. If there were resonance with lines frequency, I would see slow variations in illumination. Although I have seen that effect with conventional fluorescent lighting using video tape, it does not occur today with digital recording and either conventional or LED fluorescents. (Exposure corrections are deliberately slowed in video, to avoid reaction to page turns or other perturbations, and would highlight flicker if it were to occur.) The theoretical flicker frequency is twice that of line frequency (both halves of the cycle are used).

 

I have recorded traffic lights for legal purposes, which are LED powered under federal standards. The decay time is approximately 2 frames at 30 fps, or 1/15 second. (Cities often shorten the yellow time at intersections with red-light cameras.)

 

Line frequency is regulated over a period of several minutes or hours, and can vary significantly with variations in load on the network. This is due to inertia of the turbines or engines used to generate power. The frequency is elevated to compensate for periods of slowdown so that electric timers are accurate over a 24 hour period. LED panels, fresnels and paras are de regueur for studio video, and are specifically advertised to have no observable flicker. It is quite possible that inverters are used with a much higher frequency (e.g., 1000 Hz) to achieve flicker-free operation without the benefit of fluorescent decay.

Edited May 13, 2019 by Ed_Ingold

[rodeo_joe1]

Hmm. Well I just pointed a photodiode at a few light sources and captured the output on a digital oscilloscope.

 

Here are the traces after normalising the amplitude.

First up an ordinary 60 watt incandescent (tungsten) light bulb:

The flicker depth is much less than I expected, at about 20%. Presumably due to thermal inertia in the filament.

 

Next up, a warm white 7w LED lamp:

Whoa! A fair bit more flicker 'depth' here. About 60%, and a fair bit more than I was expecting.

 

But wait 'til you see the 5w cool-white LED:

Yowser! 90% flicker depth. I really wasn't prepared for that.

 

And lastly, the joker in the pack - a 15w CFL lamp:

Double flicker! Mains ripple superimposed on the much higher inverter frequency. I was honestly expecting this one to have the worst flicker, but it's only about 30% total.

 

So, sorry, but the LED lamps have the highest level of flicker, and from its shape you can easily see the exponential decay curve of the fluorescence.

 

PS. I suspect your cine LED panels use a DC supply internally.

Edited May 13, 2019 by rodeo_joe|1

[Ed_Ingold]

Using a slow shutter speed smooths any flicker to a negligible level. The scan time of an A7Riii sensor is about 150 msec. I would expect to see banding if there were significant flicker.

 

I've been editing video for over 8 yours today, shot under LED stage lighting at 1/60 second, 29.97 fps, with no visible flicker, even using frame-by-frame stepping.

 

Are. you sure you're not picking up EMF from the mains?

Edited May 13, 2019 by Ed_Ingold

[rodeo_joe1]

Update.

I just tested another brand of 'daylight' LED bulb, of 9 watt 806 Lumen rating.

 

With this lamp I couldn't detect the slightest amount of flicker. No point posting the 'scope trace; it's a flat line with a tiny bit of random noise superimposed.

 

Substituting the previously tested warm-white bulb showed the 60% flicker again.

 

So what's going on?

Without breaking the lamps apart, my best guess is that the 'flatline' bulb has an added smoothing capacitor fitted to its circuitry. This would make perfect sense, and add all of 10 pence to the manufacturing cost.

 

It would give the additional safety benefit of eliminating strobing for machine-shop use.

 

A smoothing capacitor would also explain the slowed light decay seen in your traffic lights Ed.

 

I calculate a capacitor value in the region of 1000 uF would be needed to sufficiently eliminate the rectified mains ripple that causes the flicker, and this would sustain some light output from the LED(s) for maybe up to 1/10th of a second after power was removed.

 

Unfortunately, there doesn't seem to be any easy way of telling flicker-free LED lamps from unsmoothed ones. The smoothed one says 'Non dimmable' on the packaging, but I believe the flickery lamp said exactly the same.

 

If anyone's interested, the driving circuit is quite simple and, working from the AC mains input, consists of a low value series capacitor to limit current, followed by a full-wave rectifier bridge (+ parallel smoothing capacitor .....or not) and into the LED chip or array of LEDs.

 

"Are. you sure you're not picking up EMF from the mains?"

 

It's not inductive pickup. That would superimpose a perfect sine wave, and what's seen is a distorted sine^2 waveform such as is got from a discharge tube driven by AC. It's definitely part of the light signal.

Edited May 14, 2019 by rodeo_joe|1

[Ed_Ingold]

It would take a very large capacitor to smooth the voltage to that extent. Active voltage regulation is more likely. LED replacement bulbs no longer have the huge heat sinks of early versions, nor do the bulbs themselves get as hot. This suggests updated technology, particularly the use of fluorescence rather than tri-color elements.

 

Banding examples seem to be limited to those huge animated LED displays at athletic events, which would would require fast response, hence low persistence.

 

ps. I would appreciate a source for a photodiode sensor, which could also be used to check shutters and other camera processes. I now have a digital scope, rescued from lab surplus.

Edited May 15, 2019 by Ed_Ingold

[rodeo_joe1]

It would take a very large capacitor to smooth the voltage to that extent. Active voltage regulation is more likely.

Not really. LEDs only take a few tens of milliamps, and a capacitor would only need to deliver that current for half a mains cycle with minimal voltage drop. Besides, no active voltage-regulator circuit can work without power. How would it 'fill in the gaps' between mains half-cycles if not for capacitive storage?

 

After my previous post, I remembered I had a dead LED lamp lying around. After crude dissection, you can see that the circuitry is minimal and all passive.

There is a capacitor in parallel with the bridge output, but its value of 3.3uF doesn't seem large enough to do much smoothing. And I have no idea what its flicker performance was like before it died.

ps. I would appreciate a source for a photodiode sensor, which could also be used to check shutters and other camera processes.

I haven't shopped for photodiodes in ages - I used to work in a research lab where they made the things. Say no more!

 

There's the practically industry standard BPW21; a so-called 'eye response' filtered PhotoDiode. It has a fairly large die and package size, which pretty much limits its use to DC photometry. Smaller die-size PDs have lower self capacitance and a faster response time; making them more suitable for shutter-timing, flash metering, etc.

 

You can use a PD passively by shunting it with a resistor to produce a voltage output from the generated photocurrent. I find that a 1 Kilohm resistor across a small PD produces a few millivolts, given a reasonably bright light stimulus, and has quite a fast response and good linearity. If you need more output and perfect linearity, the diode needs to be fed into a fast transimpedance amplifier.

See this link.

And more theoretical detail can be found here.

 

Here's a couple of pictures of a half-finished sensor for a shutter-tester.

(Excuse the awful barrel distortion on my phone camera)

 

The sensor is an SD 5620 device, which is a PD + amplifier and comparator all in the same package. The drawback being that it needs a power connection that shares a common ground with the signal output.

 

I've had the device for some while, so I'm not sure if it's still available. I'm sure there must be something similar out there though.

Edited May 17, 2019 by rodeo_joe|1

[JDMvW]

I think this is getting overcomplicated. Unless you are working to some kind of scientific standard, AWB andd other camera settings will produce decent images for casual use.

 

 

In the long run, and for any sizeable number of slides, negatives, etc., You would do best with an actual film scanner.

 

non-scanner copiers

Edited May 17, 2019 by JDMvW

[Ed_Ingold]

In the long run, and for any sizeable number of slides, negatives, etc., You would do best with an actual film scanner.

AWB is okay for "scanning" slides, but leaves a pervasive blue caste in color negative which is very hard to overcome when inverted. The colors look very washed out and distorted.

 

My motive for promoting "scanning" with a digital camera has several key points.

• Affordable, hight quality scanners (Nikon, Minolta) are no longer being made or serviced by the factory.
  
• Resolution of a 35 mm Nikon scanner (24 MP) is easily equalled or exceeded by digital cameras. If you already have a camera, the incremental expense of "scanning" is very little.
  
• Using a camera is much faster than using a film scanner. A Nikon scanner takes about 2 minutes for each frame, whereas an entire roll can be "scanned" with a camera in under 5 minutes. Before switching to digital, it would take me 3-4 days to process one evening's shooting at an event or wedding, once I had the developed film in hand. Photoprocessing took anywhere from a day to a week as mini-labs closed down.
  

[JDMvW]

I don't agree with your points.

 

leaves a pervasive blue caste in color negative which is very hard to overcome when inverted.

Any decent scanner and software package will handle the CN masking without any problem of "blue" and whatever off-color there may be is easily fixed.

• That scanners are a problem in terms of finding them is true, but if you are still working with slides, why would that worry you :)
   
  
• Increased "resolution" above 4000 ppi is illusory . It will increase the file size, but yield no additional data. The problem is not with the camera, but with the limits of the original medium.
   
  
• Setting up and individually loading and unloading a macro setup is only 'faster' if you count just the time to capture one image.
  

 

I know that you are committed to your approach, as I am to mine, but I only bring the issue up again up to suggest to those new to the topic that your approach is not the only approach.

[Ed_Ingold]

• Setting up and individually loading and unloading a macro setup is only 'faster' if you count just the time to capture one image.
  

My estimate of 5 minutes/roll is based on practical experience, and includes sorting, cleaning and loading. Copying slides requires little or no post processing. When a scanner is used, the film often picks up dust from the mechanism itself, which is nearly impossible to clean, and floating in air. When copying with a camera, the film is exposed to the environment very briefly after cleaning, and dust pickup has not been a problem.

 

Negative color film copying takes about the same time, but processing is fussy and time consuming. I use the back end of Silverfast HDR for the processing, which handles the inversion process and much of the color correction. Most of the processing time consists of cropping each frame to exclude the border areas, which have a profound effect on color conversion, and organizing a batch operation to do the processing.

 

Resolution above 4000 ppi is not entirely wasted, if you want grain-sharp results. In fact, a 24 MP sensor (6000x4000 px) is barely adequate. A Nikon scanner has 4000 ppi resolution for each color. Each cell is "exposed" successively with R, G and B LEDs, plus IR (for dust removal). Unlike with a Bayer filterLittle if any Interpolation is required in the results. Data storage is cheap at this level. I wouldn't base any decisions on this factor.

[JDMvW]

My experience is other in terms of time, and I have a Universal Repronar to mount the camera on.

 

With both my CanoScan 4000 and my NIkon Coolscan 9000, scanning a strip of slides or film is automatic, so loading and reloading each slide in a copy setup is more cumbersome.

 

As I said, my purpose was not to argue this out all over again, but simply to suggest additional methods, used by many of us.

 

I have digitized something over 94,000 film images over the last 10 years.

[Ed_Ingold]

You could buy a nice digital camera for the cost of a used Nikon roll feeder. That's on top of the scanner itself, which sells for twice its price when new. I still have my scanners, so I'm not shopping. That boat has sailed for millennials and younger.

 

Auto feed notwithstanding, a Nikon scanner still takes 2 minutes or more per frame. The Nikon is a speed-demon compared to the Canoscan 400, which takes upwards of 7 minutes per frame at high resolution. The Nikon scans up to 40 frames on a complete roll, compared to 6 frames for a strip feeder. Either way you spend a much of a day baby-sitting to scan a couple of rolls.

[rodeo_joe1]

AWB andd other camera settings will produce decent images for casual use.

Not my experience when digitally copying colour negatives.

'Removal' of the orange mask through a custom colour balance makes the whole process quicker and easier.

 

By setting a custom camera 'white' balance from an unexposed section of colour negative film, the mask can be virtually eliminated after inversion, leaving only equalisation of the RGB curves to be done. Auto White Balance can't achieve this, and in fact changes the processing needed for each negative.

 

I've found that using flash or daylight for the actual exposure, but setting the camera WB to 3400 K and + 5 Green, all but eliminates the mask for the majority of recent colour negative materials. It's more precise to set a custom balance from a blank frame of film, but that sometimes isn't possible.