The EVIL Camera Project...
Several years ago, I had an epiphany. Thankfully, I got over it by writing this article and building this model…several times…until it was out of my system.
After I posted this article on August 11, 2007, I linked to it on the dpreview.com News Forum and also on the Nikon and Sony Forums, which I was frequenting. I got lots of comments, ranging from …what in God's name is that…. Looks like it would fall apart…There are so many problems with the concept I won't bother listing those I see. Instead I'll thank you for at least keeping the idea of an compact camera innovation alive even as camera makers are too timid to offer anything fresh on their own"… to more positive comments like: "A wonderful, detailed, purposeful, and inspiring mock-up!"… "I quite like the minimalist aesthetic of the design.…That's a pretty interesting read actually.…I like the concept. I would buy one if someone like Olympus actually made one.…The concept is quite interesting.…A wonderful, detailed, purposeful, and inspiring mock-up!…Charlie, I wasn't too sure about your design the first few times I looked at it but it does actually have possibilities.……the modular design approach is a great idea, I believe.…Way to go! You even covered the dominant or missing eye problem.…This is a brilliant concept. Agree with most of your ideas…Keep the ideas flowing - it's the way we make things better. …Whatever you do with your concept, don´t abandon the eVF!…I want that camera! But…make it small, with removable battery and SD card……I do hope that some incredibly enlightened manufacturer sees the light soon and contracts you Charlie to go be their chief designer.…You put a lot of thought into this and I like the very rational and minimalist approach. I think camera makers should take note.…Charlie, this is amazing. I thought at first that you had done the "photos" in CAD, but then it appears you have built hardware mockups - keep developing this idea, I think it has a lot of promise. I'm going to have to read this through a few times and digest it before I can contribute anything of value, but offhand I really like what I see here!…I really enjoy following your design project. I, too, thought it was all CAD drawings only and am really pleased to see you have mockups for the 'feel of things.'…Very cool Charles!! You seem to have most of the technical details well in hand - the only suggestion I'd make myself would be to make a few items like that handgrip a little more organic.…Charlie, great read, lots to take in regarding rev.2 but looks very good.…I have bookmarked your page. Thank you very much for reminding over here about it, I wasn't under those who knew about your work, your excellent and fascinating work. There are many points I would like to elaborate during the next weeks, concerning your exposure of ideas, facts, ergonomics and modeling……My mom always used to tell me to 'Put down that book and go do something useful'. If I'd taken her advice, I might be more like you.…Wow - Great job and good work!…Wow Chuck, I haven't read it all yet, but you really have put lots MORE work into this camera. It looks pretty darn good! I will have to bookmark it and finish reading it at a later time, but just wanted to let you know I think this is cool. I also like that you have let your personality show thru in the commentary. Makes it more interesting and easier to read thru it. Thanks for this!…Very interesting, have you try contacting a camera maker? hopefully Nikon?…Very interesting concept. And good work! It's about time that we should do away with that archaic mirrorbox!…I dig your thinking. Generally, I don't need my digital image capture device to look like a film camera-although in my mind the rangefinder body, which is basically just a rectangle, makes sense…I would prefer the exposure time to be quoted in decimal format, but based on milliseconds as base unit (not seconds!) and rounded to 2 decimals for display in the viewfinder…What I like most about your project is that instead of just whining and complaining like the rest of us, you're doing something constructive!!!…That's a very brave page! Nice work. I hope you're prepared to deal with the fact that you're ahead of your time - many will look down on this much like they did with digital around the year 2000. And in a way, everyone is right - most eVFs today are bad, and removing the optical viewfinder can bring great rewards.…Excellent read. Looks like a very nice little camera. It's amazing that someone could actually imagine it and build it himself. It's also amazing to me that you have responded so well to feedback.…So basically you are taking all of our user opinions to make an ideal camera based solely on user input, made by a photographer for a photographer. That is excellent. Shouldn't the camera companies be hiring people like you? Seriously though, have you considered an EVIL patent? LOL. I've been following this design from the first attempt, and it's getting better with every update…Thanks for sharing! Great approach. I like your radically innovative thinking, your thoroughness and the whole spirit of your effort! And yes, I am interested in future eVF cameras along the lines you show. I have not read everything yet and certainly have not fully understood all issues involved with the concept. I would love to see your prototype materialize as finished product…I want one! NO - make that two! Fabulous work…
I'm pleased that so many people found this project interesting and generated good suggestions. If you are one of my advisors, be aware that I HAVE heard you and am trying to assemble designs that include as many good ideas as possible. Unfortunately, just because your idea was good, doesn't mean it will be used. I have to rank the ideas and some may conflict with other ideas. All ideas are good ideas, but I can't use them all. Sorry. But I will promise to consider every suggestion…
This article was originally on my blog, but it became defective and riddled with SPAM submissions, so I pulled it down. I'm posting it again on my website.
For those that don't know, "EVIL" stands for "Electronic Viewfinder with Interchangeable Lens". It's the mythical Holy Grail for many people, combining the eVF from one type of camera and the Interchangeable Lens from another type. In that sense, it's the ultimate "Bridge" camera! After many years of predicting that an EVIL design was pending (and not being a good futurist), I decided to explore what an EVIL Bridge camera might look like and how it might work. While I'm at it, I decided to address several gripes I have with the design of most cameras. Note that this predated the 1st real EVIL camera, the Panasonic G-1…my concept is quite different than what the G-1 designers came up with.
Here were my major gripes with existing cameras:
1. I know how SLR cameras are built and appreciate why the viewfinder must be centered from an opto-mechanical perspective. But far too many cameras that don't need the viewfinder there, conform to the "SLR look". I always wanted a viewfinder on the left end of a camera body, where my nose could lie to the side and not get "smushed" up. With an SLR, it was just a comfort issue, however, with a modern electronic camera, there is normally an LCD right where my nose lands. On my R1, which has the LCD on the top, there are buttons where my nose resides. Duh! I can't imagine why everyone insists on centering the eVF!
2. I have big hands. I have never found a truly comfortable camera design. The R1 comes close, but even it is too small (it has a huge "hand grip" area in comparison to most cameras). My test for a proper sized hand grip is that I must be able to comfortably operate the camera with one hand. When I try this with most cameras, I find that the camera wants to fall as I use my thumb and index finger to press buttons. I want a place for my right thumb to wrap around the camera…it's really hard to hold a camera with only my other 4 fingers (there is a reason natural selection gave us one opposing finger)!
3. Lousy camera control system. I've seen a few innovative cameras lately, but nobody does one like I want. The best of the current crop has about 20 dedicated buttons and about 10 levels of menus. This makes these cameras confusing and error prone (there are few places you can grab the camera w/o danger of pressing a button). Combine this with the Japanese tendency to miniaturize everything and you basically get a camera totally covered with little buttons!
4. Cameras tend to be too small. This is similar to my 2nd and 3rd gripe, but different. I laugh when I hear young, urban whippersnappers complain that a camera is so big and heavy that carrying it all day is impossible. Yet, they have health club memberships and use them. I'd think holding a heavy camera was similar to lifting free weights? Wimps! A big camera is easy to hold steady…and it's comfortable. A heavy camera doesn't need IS/VR…much.
5. There are some traditional camera nomenclatures that need to disappear! While I thought long and hard about dropping the f/N convention of expressing aperture (which confuses many), but it is simply too logical to go away. But the "1/N" way of expressing the exposure time should go away and be replaced by a decimal fraction. For example, an exposure time of 1/8000 second would become an exposure time of 0.13 milli-seconds. Often the status quo leads to dropping the "1/N" convention, being replaced by just "N", which can be quite confusing...like what does "shutter speed of 100" mean? Many beginners struggle with this! I'll also lead the way by never again using the term "shutter speed", as "exposure time" is so much better and uses exactly the same number of characters. I'll also drop the term "ISO" when describing the sensitivity…ISO has thousands of specifications that describe more than just camera sensitivity. "Sensitivity" will work quite well and "Sen" will work when space is limited.
I've tried to address all these gripes in my design goals:
o Mount the optional viewfinder on the end
o Make the optional hand grip big (supply 2 different size covers)
o Minimize the buttons and depth of menus
o Make the camera properly big (but not TOO big)
o Consider people with dominant left hands and left eyes (& missing eyes)
o Design for people with different size hands
o Use a large, touch-screen LCD
o Make the primary lenses short-back-focus designs
o Provide an adapter that will accept legacy long-back-focus lenses
o Make the interface between the camera and optional flash FAST
As a start, here are specifications for the EVIL Camera System:
* Sensor vibrates in Z-axis to determine direction and approximate distance of focus, with a slew to that position, followed by conventional iterative contrast AF.
** AF Assist happens automatically when LED Flash is mounted. A single light pulse occurs; the camera performs AF Assist at the beginning of illumination.
I am NOT a rabid Apple fan. I use a PC, like most "normal" people. Since I have an iPhone, I just appreciate the industrial and human factors design that goes into Apple products.
The human factors design of The iPhone it is brilliant. I love the concept of "gestures" and incorporated them in my design. I don't think my design would violate Apple's patents, but if it does, a licensing fee is appropriate.
The LCD on the back is the same size as the one on the iPhone. I've tried an iPhone in bright sun and found it quite usable. It's big enough that you can use your finger, instead of a plastic stylus (thus, the screen won't get scratched…besides, it's glass instead of plastic).
I decided to use a 1.5 crop factor sensor, but allow the camera to reduce the resolution when a lens for a smaller format is mounted. I've outlined 8 different formats as this table shows:
This approach keeps the photosite area fixed at about 4 square microns, which is a good compromise between resolution and noise. This size photosite is found in several common dSLR cameras, such as the Nikon D80, Pentax K10D, and Sony A100. The Canon 30D has a slightly larger photosite and the 40D has a slightly smaller photosite. Note that I have defined two crop-factor 2.0 formats…one for the 4:3 System standard and one for a 3:2 aspect ratio. I've also defined a 16:9 ratio. I know that a multiplicity of formats may be seen as a gimmick, but if you are using a Zukio lens that only covers an 18mm x 13.5mm area, it seems silly to scan and save the entire 24mm x 16mm sensor…that just wastes 60% more memory and 60% more time. By doing this, I also get rid of the "digital zoom" concept. It makes more sense this way…it's not really a "zoom" function when you "crop"…
There are diagrams of these 8 formats farther down…
With a 1.5 crop-factor and 10 MP, the photosites are relatively small and have small noise and dynamic range issues. I dislike the trend to apply heavy NR in-camera, as the damage to detail can't be reversed.
I provided for cooling in the design which should improve the noise level. The next thing that needs to be improved is dynamic range (DR). There are many ways to do this:
o Larger photosites (and larger electron wells associated with them)
o Lower noise floor (cooling the sensor and electronics)
o Dual photosites (one big…one small)
o Novel ways to count photons
For a moment, I want to consider that last one…
For several years, proposals to improve the DR of photosensors have resulted in a flood of patents. These are intriguing concepts which promise to have dramatically better DR. I have seen results claiming a DR (or signal-to-noise ratio) of 136 db. To put that into perspective, the best photographic sensors today have a SNR of about 70 db (136 db is over 10X better than 70 db). Let's think how these ideas do that…
There are many proposals: Logarithmic sensors, "clipped" sensors, multimode sensors, frequency-based sensors, and sensors with control over integration time, to name a few.
In general, these all eliminate or reduce overflowing of the capacitor ("electron well") associated with each photosite. When the well fills, additional electrons kicked out by photons have no place to be collected and the detector overflows. At this point, the sensitivity truncates and limits the range. When these "extra" electrons escape, they can flow into surrounding photosites, where they cause "blooming".
The solutions for this generally reduce overflow by detecting when the well is nearly full and discharging the capacitance. This, in its simplest form, converts the output of each photosite to a variable frequency, proportional to the incident illumination. A counter accumulates this variable frequency output during the time of exposure to the image. In theory, the size of the electron well no longer limits the DR. Instead, a smaller accumulation well will produce a higher frequency. The new limit on DR is the size of the counter. Overflow now occurs in the digital realm instead of the analog realm in current sensors. The combination of the detector which resets the accumulation capacitor prior to overflow and the circuit which counts the number of times that the accumulation capacitor is reset, comprise an analog-to-digital converter.
Most of the recent patent activity seem to be small tweaks to this basic invention and/or attempts to circumvent it.
A successful implementation of this concept seems close at hand. It will be a MAJOR improvement! It will increase the IQ of sensors with tiny photosites. This, I think will validate my choice of a reasonably small sensor for this project.
The EVIL Camera is a "system". Not only does it have a removable lens, but also a removable hand grip and removable viewfinder. There are 2 covers for the hand grip that should accommodate most hand sizes. I mounted the flash above the hand grip so that the weight of the flash doesn't contribute to rotational torque, making the camera + flash easier to use. The body can be replaced/supplemented with a different version (bigger buffer, more or less pixels, faster processor, B&W version, etc). The same lenses, hand grip, and viewfinder will work with all the bodies. Since the sensor is close to the front of the body and there is no shutter or mirror in the way, the anti-alias/hot mirror filter can be easily cleaned, thus not much need for auto cleaning (and it's mostly a marketing gimmick anyway). The hot mirror filter can also be easily removed and replaced with an IR pass filter.
The adapter for legacy, long-back-focus lenses will look simple…just a tube with appropriate mounting and electrical contacts on the ends. Inside will be complex, as there must be (in some instances) a focus motor, a shutter, and electronics to interface between the camera and lens. One reason for using a short-back-focus system is that it will allow the use of almost any 35mm SLR or dSLR lenses!
However, having to include a shutter in the lens adapter is a difficult one. There is not much room, which will necessitate using a multi-blade circular design. It also will not be located in the best optical location...there will be some issues with higher exposure times.
I chose to NOT put an AF motor in the body, thus, to get AF will require a lens with an AF motor. I have specified all the lenses designed for the EVIL Camera to have AF motors. For the other lenses, they will all require an adapter, which can have a focus motor for lenses that don't have one. I think it should also be possible to use manual lenses, by telling the camera the max aperture and FL…and keeping the AF active during manual focus operations.
Since there are multiple configurations, I decided to put a small battery in the body and a big battery in the hand grip(s). Both will be replaceable. Both the body and the hand grip will have replaceable flash memory.
If you're still reading, you deserve to know more about what this camera design looks like on paper (it's basically the same information in the Specifications above, but not so structured and more readable...I hope):
o Body is 4 x 3 x 1.5""
o Has 24mm x 16mm CMOS sensor, with 10.1 MP, but other resolutions down to 5.7 MP are selectable
o The sensor has a native 3:2 aspect ratio, but some resolutions have4:3 and 16:9 ratios
o Sensor is cooled by fins on rear of body (under LCD)
o Cast Mg frame construction, with cast Al back (for thermal properties)
o Optional hand grip…will connect on right side of body via magnets
o Hand grip comes with 2 add-on covers to expand the girth (making 3 different sizes)
o Connector on top of hand grip mounts flash (and keeps weight centered above right hand)
o Plastic construction, w/ metal frame
o Optional eVF…will connect to left side of body
o eVF mounted at extreme side so that nose will be at side of body
o eVF rotates 90 degrees (up) so that right eye can be used
o eVF resolution is 480 X 640 pixels (921,600 total px; 640 ppi)
o Adjustable diopter correction (-2 to +1)
o Plastic construction
o LCD centered on back of body
o LCD rotates 270 degrees
o LCD twists 180 degrees
o LCD 2" X 3" (same size as on iPhone)
o LCD resolution is 480 X 640 pixels (921,600 total px; 240 ppi)
o LCD touch sensitive
o LCD illumination automatically increases in high ambient light
o Optional viewing hood clips on LCD (thus functions as a VF)
o Most controls are via touch screen
o Minimum buttons/controls on body/hand grip
o Orientation sensor rotates text/menus (but not pictures)
o Orientation sensor also tags pix in EXIF
o Twist/flip sensors orient information on LCD properly
o Plastic construction, w/ glass LCD cover
o Two LiIon batteries…one in body…one in hand grip
o Battery in body is small and replaceable (door on left side)
o Battery in hand grip is big and replaceable (door in bottom of hand grip)
o Battery in flash is LiPoly type
o Batteries in cooler are either alkaline, LiMH, or Lithium AA cells
o Recharger accepts all batteries (LiIon, body/hand grip and LiMH, cooler)
o Lens centered in camera body (symmetrically, L-R)
o 5 lenses designed for the EVIL System:
o 20mm (30/40) f:2.8 lens
o 55mm (83/110) f:3.5 macro lens
o 35mm (53/70) f:1.8 lens
o 16-80mm (24-120/32-160) f:2.8-3.5 zoom lens
o 150mm f:2.8 lens
o All above lenses are short-back-focus type and have internal SSM
o All above lenses have fixed rear element to preclude "pumping" during focus/zoom
o All above lenses are sealed
o Body has large, fast cache memory for burst shooting at 5 fps for 25 frames (@ full rez)
o Both the body and hand grip have SD slots
o Selectable modes to use body SD memory, hand grip SD memory, or both (in sequence)
o Body can operate w/o eVF and hand grip
o Live preview via BOTH eVF and LCD
o Advanced contrast AF system *
o Camera communicates quickly with TTL flash (USB 2.0)
o Graduated sensitivity (varies across pic; angle and rate variable)
o Downloaded shutter sounds (for those who don't like a quiet camera)
o Tripod mount is in center of bottom
o When hand grip mounted, the button on body can be reassigned
o Optional sensor cooler attaches to back of camera (with LCD flipped up)
o Cooler is self powered by 4 AA cells…cooler uses a small TED (Pelletier effect)
* The sensor moves in/out to give info about how much/which direction to move the lens. The AF system can detect horizontal, vertical, or both features, unlike most Contrast AF designs.
Legacy Lens Adapter…
o 3 models for: 1) FF 35mm lenses and 2) 1.5-1.6 Crop Factor lenses & 3) LBF 4/3 lenses
o Front bayonet interchangeable for Nikon, Canon, Minolta, Pentax, & Olympus lenses.
o Has "fly-by-wire" ring to focus lenses that don't have external focus ring.
o Has focus motor for lenses that don't have internal AF motor
o Has stop-down lever
With only 5 buttons, 1 wheel, and 1 joy-stick, the vast majority of the controls are on the LCD touch-screen. The big button on the hand grip is the shutter button, of course. The three small buttons on the hand grip are programmable...as is the single button on the Body (when it's not working as the shutter release). I think it would be useful to select what shows on the LCD, ie, either the live view or the recorded view. The thumbwheel is intended to control the exposure. In P Mode, it controls both the shutter and aperture; In A Mode, only the aperture; in S Mode, only the shutter; in M Mode, both (the thumb-wheel also has a button, like on a wheel-mouse, which toggles between shutter control and aperture control).
The button on the body can also be programmed. When the hand grip is not used, it will be the shutter button. When the hand grip is used, this button is user programmable.
The LCD touch-screen does most of the control. It is laid out like this:
The controls on the right side are: Cursors, Burst Setup (3 shots at 4/sec), Histogram, Flash (on), File Format (JPEG), Timer (3 seconds), White Balance (sun), & Resolution (Crop=1.5).
The controls on the bottom are: Focal Length (24mm) & Macro/Distant Icons (Macro), Focus Distance & limits (47 ft, 37 ft to 1000 ft), Sensitivity (100 ISO) and Battery Status (320), Exposure Time (0.002 mS)Aperture (f/5.6)FOV (43 degrees), EV Scale (-0.75), and Memory Usage (46 of 197 shots left).
Most touch-screen controls work the same way. They BOTH display settings and also allow control of the options. As a first example, the Histogram control shows the actual histogram of the scene in the Live View section of the LCD. Press the Histogram control once and it switches from a composite histogram to a RGB histogram. Press it again and it returns to the composite (luminance) view. Press the Histogram control and drag toward the center of the LCD screen and the Live View is replaced with a larger version of the histogram. Press the Histogram control and the histogram becomes partly transparent, allowing the Live View and the histogram to both be seen simultaneously. Drag toward the lower-right corner and the controls disappear and the histogram/Live View are shown using the entire screen. Drag again in the up-left direction and the controls reappear.
Some controls have no or little actual control. For example, the FL area does not control the FL…it just reports what the FL is. This is also true of the Battery Status and Memory Usage areas.
However, some of the controls have unusual properties. Take for example, the Exposure Time/Aperture/FOV control. In addition to displaying the current exposure time and aperture, it also controls the exposure mode. Although I described the 4 exposure modes as P, A, S, and M, these letters are not actually shown. In P Mode, both the shutter and aperture are shown in black. Upon pressing it once, the aperture is shown in red, signifying A Mode. Press it again and the exposure time will be red, signifying S Mode. Press it again and both the aperture and exposure time will be red, signifying M Mode. One more press and it returns to all black…P Mode. When in A, S, or P modes, the EV Scale can be used to vary the exposure…just slide your finger back and forth. Alternately, slide your finger on the picture area; up/dn controls the aperture and R/L controls the exposure time.
The thumb wheel also can be used to control the EV. It works in the same way as dragging across the EV control, except it rotates...CW lowers the exposure...CCW raises the exposure.
There is an optional Help feature, selectable by a menu choice. Once enabled, when you press on a control, a short description will appear over the Live View area. For example, while pressing the Flash control, not only will the icon change to indicate the flash mode selected, but if the Help feature is turned on, a description of the flash mode, such as "Flash Off" or "Flash Auto", will briefly appear over the Live View area. Additionally, a description of how to use the control will appear, such as, "Tap this control to turn the flash on/off. Drag the control to the center to access a control screen for the flash system."
Continuing this example, the Flash Control Screen looks like:
The Cursor control needs special explanation. When the "+" is pressed, a "+" cursor appears over the Live View area. This cursor can be moved by sliding your finger over the screen, dragging the cursor. The cursor can be used to position where the exposure readings are taken. Even though there are the usual range of exposure options, all of them can be relocated by moving the cursor. For example, suppose the entire scene is used to set the exposure…by placing and moving the "+" cursor down a bit, the exposure weighting will be shifted so that the lower part of the scene has more affect on the exposure.
The "" cursor controls the focus location. It works in the same manner as the Focus Cursor.
Tapping twice on the Live View area or dragging the cursors back to the top-right will remove them from the screen and lock-in the settings (the locations still are effective even though the cursors are on the LCD Monitor...putting it away simply keeps the location from being accidently changed). However, you may want to leave the focus cursor active?
Note that near the lower-left corner (in the focus distance control box), is a green dot to signify when AF has found focus. This green dot is persistent (when the rest of the controls are covered by a large Live View, the focus confirmation dot remains).
This lengthy description does not fully describe the EVIL Camera System controls. It should give you a general idea of the possibilities.
Now some pictures of the camera models…
The LCD can be flipped over to protect it.
Or flipped up. In this view I have also shown the optional Grip and Viewfinder, plus a Nikkor lens adapter (Nikkor, because I have LOTS of them to use as props).
Or flipped forward, where it works much like the LCD on an R1. Note also that the eVF can be rotated through 90 degrees independently of the LCD. I've attached a 135mm lens.
When the LCD is pulled out from its stowed position in the body, the cooling fins on the back of the body are exposed to ambient air. Thus, it should be normal to pull the LCD Monitor out.
The optional sensor cooler can be mounted on the rear of the body. It's held there by magnets. The temperature of the sensor is recorded in the EXIF data and can be viewed on the LCD.
The 3 buttons on the rear are programmable. Notice also that they "feel" different. The center one is flat, the top one protrudes, and the bottom one is depressed. This way, you don't have to look at the back to be sure your finger is on the right button. The dial/joystick has several functions. The outer part rotates, which will be used to vary the exposure. The joystick has 5 switches (Up, Dn, R, L, In). The In switch signifies "do it". The joystick is an alternate to using the touch screen, (when it's not in the body, it may be difficult to use). With the cursor visible, use the joystick to position it over a control area and then depress the joystick…this is the same as tapping the screen.
The sensor is mounted near the front, where it can be easily cleaned.
This sensor has 8 different formats…six 3:2 formats, one 4:3 format (the same as a "4/3's" camera and all small cameras), and one 16:9 format. These two diagrams show the relationships of these 8 formats:
This picture shows an adaptor for Nikkor lenses mounted to the body. Note also that the eVF is reversed, which reduces the width of the camera assembly for storage. It's held in place by a small magnet.
This picture shows the two magnets that affix the hand grip to the body. Yes, it REALLY holds it securely! The magnets I used are so strong, I almost can't get it off!
Two identical magnets hold the cooler on the back of the body. It's easier to remove than the hand grip.
When the hand grip is mounted, the two shutter buttons are positioned to be operated by the right index and middle finger. The button on the body is programmable. I think I would have it toggle the preview mode…press the shutter with your index finger and then w/o moving your hand, press the button on the body to look at the picture you just took…but that's just me.
With the 20mm lens mounted and w/o the Grip or eVF, the camera is quite compact.
As this picture shows, my big hands CAN with no difficulty hold the camera body w/o the Grip. Note that the shutter is operated by my middle finger.
This diagram shows the major components:
To show relative size, here is the EVIL Camera with a 135mm lens vs one of my Nikon F bodies and a 300mm lens:
Note that in the above comparison, if the 135mm lens was designed for the EVIL system (ie, short-back-focus), it would be smaller...plus there would be no adapter required.
Let me comment briefly on some of the most radical aspects of the design.
I didn't specify IS/VR in the body. I do have the sensor moving, but in the Z-axis. I don't think IS/VR is as important as fast AF, so I had to make a choice. Allowing the sensor to vibrate in the Z-axis will allow the AF system to initially estimate where the correct focus point is (in much the same way that phase AF systems do…but differently) and then make a big focus adjustment, followed by a small contrast adjustment. This approach should be both fast and accurate.
I've designed a special flash unit for the camera. It had to be narrow enough on the left so that the LCD doesn't hit it when swung forward. I designed the flash to have a bigger source than the typical flash, to minimize shadows w/o having to use a diffuser.
After some research (ie, reading the literature), I concluded that it's time for a LED-based "flash" unit to appear. The reason it's time is that LEDs are maturing and specifically, large, high-power GaN/InGaN devices are becoming available with very high efficiency. A good read about this is at:
Here is a quote:
"In September 2003 a new type of blue LED was demonstrated by the company Cree, Inc. to give 24 mW at 20 mA. This produced a commercially packaged white light giving 65 lumens per watt at 20 mA, becoming the brightest white LED commercially available at the time, and over four times more efficient than standard incandescents. In 2006 they demonstrated a prototype with a record white LED luminous efficacy of 131 lm/W at 20 mA. Also Seoul Semiconductor has plans for 135 lm/W by 2007 and 145 lm/W by 2008, which would be approaching an order of magnitude improvement over standard incandescents and better even than standard fluorescents. Nichia Corp. has developed a white light LED with luminous efficacy of 150 lm/W at a forward current of 20 mA."
It should be noted that high-power (> 1 Watt) LEDs are necessary for practical general lighting applications. Typical operating currents for these devices begin at 350 mA. The highest efficiency high-power white LED is claimed by Philips Lumileds Lighting Co. with a luminous efficacy of 115 lm/W (350 mA)."
Most white LEDs in production today are based on an InGaN-GaN structure and emit blue light of wavelengths between 450 nm and 470 nm. These GaN-based, InGaN-active-layer LEDs are covered by a yellowish phosphor coating usually made of cerium-doped yttrium aluminum garnet (Ce3+:YAG) crystals which have been powdered and bound in a type of viscous adhesive. The LED chip emits blue light, part of which is efficiently converted to a broad spectrum centered at about 580 nm (yellow) by the Ce3+:YAG. The single crystal form of Ce3+:YAG is actually considered a scintillator rather than a phosphor. Since yellow light stimulates the red and green receptors of the eye, the resulting mix of blue and yellow light gives the appearance of white, the resulting shade often called 'lunar white'."
White LEDs can also be made by coating near ultraviolet (NUV) emitting LEDs with a mixture of high efficiency europium-based red and blue emitting phosphors plus green emitting copper and aluminum doped zinc sulfide (ZnS:Cu, Al). This is a method analogous to the way fluorescent lamps work. However the ultraviolet light causes photo-degradation to the epoxy resin and many other materials used in LED packaging, causing manufacturing challenges and shorter lifetimes. This method is less efficient than the blue LED with YAG:Ce phosphor, as the Stokes shift is larger and more energy is therefore converted to heat, but yields light with better spectral characteristics, which render color better. Due to the higher radiative output of the ultraviolet LEDs than of the blue ones, both approaches offer comparable brightness."
The newest method used to produce white light LEDs uses no phosphors at all and is based on homo-epitaxially grown zinc selenide (ZnSe) on a ZnSe substrate which simultaneously emits blue light from its active region and yellow light from the substrate."
A new technique developed by Michael Bowers, a graduate student at Vanderbilt University in Nashville, involves coating a blue LED with quantum dots that glow white in response to the blue light from the LED. This technique produces a warm, yellowish-white light similar to that produced by incandescent bulbs."
In the last couple of years, LED-based automotive headlights have appeared.
Some cell phones have LED illumination for their cameras. Several companies are aggressively marketing 1-2W LEDs to cell phone manufacturers. Here is an interesting article by CAP-XX, an Australian supplier of super-capacitors:
They suggest that 2-4 of the LumiLEDs LXCL-PWF1 be used for a high-end camera phone.
Another informative article:
The most common "white" LED is a blue LED with a yellow-orange phosphor on the face. It produces a spectrum that appears white to the human eye, but is weak in the red & green wavelengths. It would not be the best choice for color photography. Better solutions would be a matrix of R-G-B LEDs, a blue LED with a red & green phosphor, or a matrix of a blue LED with an orange phosphor plus blue-green and red LEDs. I personally think the latter solution is the better one, as it has a reasonably even color spectrum and allows the color temp to be varied.
There appear to be several sources of 7mm square LED assemblies, designed to surface mount on PCBs. In this size package, 10W to 20W outputs are available. Pulse power is generally 3-4 times the steady-state power.
Taking conservative estimates, ie, 10W * 3 = 30 watt pulse power per LED module and 50 lm/W, I have packaged 35 modules on a 2.2" x 2.8" (55.9 x 71.1 mm) assembly. That translates to 60,000 Lumens! My second approach packages 100 LEDs on two 2.75" x 4.15" (69.9 x 105.4 mm) assemblies. Assuming a mix of 10W (peak) white, teal, and blue LEDs, that translates to 1000W peak output. Assuming again 50 lm/W, that's 50,000 Lumens! And with a luminous surface area of 23 square inches, it should produce soft shadows.
A LED design has some good operational characteristics:
o Variable Power Output by changing pulse current
o Variable Energy Output by changing pulse duration
o Flash duration can equal exposure time, if desired
o Multi-pulse mode possible
o Quick "Recharging" since storage capacitor is not drained
o Broad source minimizes shadows
o Variable color temperature, by varying current/duration of individual LEDs
o Lower nominal color temperature than 6500 degrees K
o Silent (no DC-to-DC converter)
o Efficient (no DC-to-DC converter)
o No high voltages
o Essentially zero standby current
A LED design also has some operational disadvantages:
o Not possible to stop action as well as with Xenon arc tube
o LEDs get hot
o As LEDs get hot, output drops
o LEDs can be destroyed by high temperatures
o Expensive (at this juncture)
There are some technical challenges…
At a typical forward voltage of 3.5V, these super high-power LEDs will require quite high current…about 2.8A per diode. Add that up and you get about 100 Amps for an array of 35 LEDs! If some LEDs are connected in series to utilize a higher voltage, for example 3 LEDs in series will require 10.5V, the current will still be high…33 Amps in this example. With the 100 LED design, the currents are a bit lower, but still high. This will require a VERY "stiff" electrical source…something like Lithium polymer batteries, perhaps in parallel with a big super-capacitor.
The transistor switch(es) will have to be BIG. Having several different color LEDs in the array helps some by requiring separate driver transistors, thus sharing the current load. It's probably wise to anticipate several drivers anyway.
This LED flash model is quite small, compared to other flashes with comparable output.
I received a comment on dpreview from Joseph Wisniewski about the LED Flash concept:
"On your LED flash, you'll need at least 8, and preferably 10 or 12 colors of LED. (I've designed abridged spectro-colorimeters that used up to 12 LEDs). LEDs are near monochromatic: an object that reflects primarily orange light looks orange under white light, and black under red + green + blue light."
I have done some thinking about Joseph's suggestion. Joe is one of the most intelligent guys I know and I don't disregard his advice easily. Since a sensor only employs 3 colors of filters, I doubt that I'll need 8-12 colors of LEDs. What Joseph was doing was much more critical than a camera. Here is how I came to that conclusion...
Here is a curve of the outputs of a "white" LED and 4-monochromatic LEDs:
When they are combined, they produce a composite power curve:
This is not perfect and as Joseph suggests, having 8-12 different wavelength LEDs would allow more uniformity. But while this curve might not be perfect, is it good enough?
Before we go down that path, let's look at the other end of the system. Most Bayer sensors have 3-color filters over the photosites. These have a typical response of:
When combined, they give a composite sensitivity curve for the sensor:
Part of this curve is simply the natural sensitivity of Silicon:
The rest of the sensor response curve is controlled by the RGB filters.
Finally, what is the spectral output of a Xenon arc tube? Well, first, it's variable. If the current density is low, the output is a bunch of narrow spectral lines and is not too suitable for photography. If the current density is high, it's dominated by "black body" radiation, but with bumps due to the spectral nature of the Xenon. Here is the curve from a high-current-density tube that is unfiltered:
Many commercial photographic flash units have filters that suppress the UV output past 400 nm. I think they do this because UV can be destructive to paint and ink. That's one advantage of the LED approach: LEDs don't emit UV (unless special UV LEDs are used).
My feeling at this juncture is that all light sources have characteristics that make them non-perfect for photography. That is one reason why cameras have a white balance feature. My assumption is that creating a new light source, with a different spectral curve is OK. Thus, while I understand Joseph's concern, I think that a quite adequate LED flash can be done with 1/3 to 1/2 the number of different color LEDs than he suggested.
One of the nice features of this LED flash (one which I just mentioned briefly) is the ability to vary the color temperature. One problem with Xenon arc flashes is the very blue spectrum. This is not an issue when it's used as fill for a daylight shot, but when used in relatively high ambient artificial light settings, the mixture of colors is a problem. With an LED flash, the light can be balanced to match the ambient w/o the use of filters.
Just as with a custom WB, the LED flash can be balanced. I'd propose that if the LED flash is attached and enabled, the camera should balance the flash to the ambient as a default. Even when the camera is set to "Auto WB", the LED flash should mimic what the camera determines the ambient color temperature is. Put another way, the LED Flash can use the camera's Auto-WB information to produce a spectrum which matches the light that the camera sees.
I think TTL flash needs a new paradigm badly. With a mirror camera, the TTL pre-flash has to be followed by a delay, waiting for the mirror to flip up. Even some non-mirror cameras have a major delay between the pre-flash and the main flash. This delay is terrible for a couple of reasons: 1) the subject moves between the full-press and the actual exposure and 2) the subject blinks and changes expression. This big delay seems worse with external flashes, probably due to the slow serial interfaces between camera and flash.
The EVIL Camera doesn't have a mirror, so the pre-flash can be prepended to the main flash:
1. The flash turns on
2. The sensor samples lighting (where the photographer said to take the exposure)
3. The camera calculates the power level and duration (based on the flash settings)
4. Clear the sensor (with a CMOS sensor, this is instantaneous)
5. Start accumulating the image
6. When the calculated time is up, shut the flash off
7. Continue accumulating the image until exposure duration is satisfied
8. Alternately, the flash can remain on until the exposure time is up
Note that there is only ONE flash and capturing the picture starts a few mS after the flash comes on. Another wrinkle is that the flash can turn on during AF, remain on during exposure calculation, and remain on through the image exposure…there is no need for a separate AF assist LED or laser. When using the LED flash as an AF assist (but not using the flash), it will blink.
I've designed and built several LED flash models. Each one improved on the concept, I think. At the last iteration, I revised the hand grip. I added a "groove" on the front for my index finger and softened the curves a bit. The reason I was revising the hand grip is that I removed the conventional "hot shoe" and created a radically different means of attachment for the flash: A rectangular hole in the top. The hole has a trap-door to keep out trash. The trap-door is hinged on one side and when you push the flash down, it swings to one side (out of the way). The electrical contacts are on the opposite side of the well from the trap-door. When the camera applies power to the flash, a small motor operates two clamp fingers on the front and back of the foot on the flash. These clamps hold the flash in place until the release button is pressed.
This view shows the new "trap-door" for mounting the flash on top of the hand grip.
Rear view of the Flash, with LED "wings" folded:
A view of the LED "wings" folded forward (note that they don't fold out all the way where they are perpendicular to the flash body…they point slightly to the sides to distribute the light more uniformly). Although with longer FL lenses, a perpendicular detent could provide a bit more light:
To bounce the light, the entire LED "wing" assembly swings up:
Now that I've got the LED flash models pretty much completed, I want to go back and think about the sensor.
OK…back to the camera. Although I would like to be smart enough to anticipate everything and say that all neat consequences were anticipated, I'll also accept accidents when they happen. Originally, I set out a goal of having an eVF that could be used by either eye, but abandoned that in the interest of practicality. I was surprised a few days ago when I was playing with the model and discovered that if the eVF was flipped up, it was easy to use my left eye, w/o my nose hitting anything!
Here are a couple of pictures showing the two ways to use the eVF:
I hope this EVIL Camera Project has been interesting and informative. If you have constructive comments, please send me an e-mail.