How-To: Ghetto Flo

Executive summary on Home Gown Flo lighting, A.K.A. Ghetto Flos:

• The color quality is a result of the tubes only. ( The technology that makes the light is in the tube. )


• Modern high frequency electronic ballasts (which are very common now) do not flicker.


• There are many good choices of off-the-shelf tubes with the **CCI and a high *CRI.


• It is possible to "over-drive" off-the-shelf tubes with off-the-shelf ballasts to get higher output.


• Compact Fluorescent bulbs are a very promising technology for Ghetto-Flo lighting. They use electronic ballasts and have high output for their size and are available in the correct **CCI and CRI.

*CRI - Color Rendition Index. How continuous or "true" the spectrum from a light is on a scale of 1-100. 100 is perfect. Lights with a CRI over 85 are suitable for film and video. A CRI of 90 or better is great, but the light output of extremely high CRI lights is often lower than lights in the area of 85 CRI. Under 80 CRI is not good for film and video.

**CCI - Correlated Color Index A.K.A. "Color Temperature" - The theoretical black body color value of the light.

A quick survey of the available off-the-shelf hardware reveals that good home-brew flo lights can be built for significantly less than the cost of name brand flo-lights. (Anywhere from 1/20th the price to 1/6th the price, depending on various factors)

Intro:

The problem:
I needed new lights to replace my old photo light setup. The new lights had to meet certain requirements.

1. Must work for both still photography and film/HD/Video
2. Capable of producing soft light without taking up too much space.
3. Daylight balanced so they can be used in combination with natural light
4. As inexpensive as possible.

Kino-Flo (TM) and various other name brand manufacturers produce movie lights based on fluorescent tubes rather than tungsten or HMI bulbs. As you might expect, they emit a soft wrappy light similar to a soft box of the same size. Kino-Flo Tubes are available color balanced at 5500k (Daylight/HMI), 3200k (Pro-Tungsten) and 2800k (Consumer Tungsten). Kino tubes are engineered to have a good Color Rendition Index (high spectral continuity) and lack the typical green/blue spike many consumer and industrial fluorescent tubes have. Beyond the precision tubes, another "special" thing about Kino-Flos is the ballast. Kino ballasts are engineered to be flicker-free when shooting at normal motion picture frame rates.

For soft light, Kino-Flos kick are one of the finest solutions available. They have high output for the amount of power they use (nearly three times as efficient as Tungsten lights) and they hardly generate any heat. The problem is Kino-Flos cost too much. Even the tubes cost too much. ($20 each)

Would it be possible to build something equal in caliber for less money?

Theory:

I'd read that others had been experimenting with some of the (new-ish to the consumer market) electronic ballasts. In theory, almost any modern electronic ballast should be stable enough to be flicker free at regular film frame rates. While doing my research I noticed that there is now a commodity consumer market for "Continuous Spectrum" fluorescent tubes as well. In combination, these two consumer grade off-the-shelf technologies should be capable of producing a quite serviceable "Poor-Mans Flo" light. The name that seems to have caught on for this type of lighting instrument is Ghetto-Flo. From here forward I'll refer to them as either Ghetto-Flos, G-Flos or home-brew.

Testing Begins:

Though I've read a few accounts of peoples success building and using Ghetto-Flos, the accounts always seemed a little thin on details. Not a big problem really. But it did take me a while to gather all the information I needed. I tend to not really believe anything until I see it myself so I decided to do a series of tests to figure out my own "recipe". Considering that the bulk of the "technology" in a flo-light resides in the tube, the thing I was most interested in learning from my tests was what the quality of the light from various commercially available tubes looked like. Second, I was interested in the stability of the ballasts when shooting various shutter speeds/frame rates. Last but not least, I was interested in the total light output (LUX) from the ballast / tube combination.

While I attempted to be as methodical as possible my testing method is not highly scientific by any means. I do not have a color temperature meter or the equipment to accurately verify manufacturers published CRI, CCI or relative output of the various tubes I tested so I came up with a simple method to at least approximate the quality of each tube. I shoot a test scene with my digital camera in "RAW" and then color correct it for white balance in Photoshop RAW loader to determine it's approximate white point. I then color correct the same image to the stock "Daylight" and "Tungsten" white points to see, visually, how much the light color deviates from that white point. I do both tungsten and daylight because even though my current focus is daylight balanced tubes I may eventually test tungsten color balanced tubes. For the relative light output, I just use the exposure information from the EXIF data of the image which tells me if the light output is significantly different from one setup to another.

Here are the results from the tubes I've tested so far:


Recipes:

There are several recipes for home-brew flo lights. Here are a few I've investigated.
1. Use and off-the-shelf lamp housing with an electronic ballast + the correct flo tubes
2. Fabricate a lamp housing or modify one and use off-the-shelf electronic ballast (popular when over driving tubes)
3. Compact Fluorescent "bulbs" with reflectors and/or "China Balls"

Recipe #1 ingredients:
2 x GE Chroma 50 T12 48" [5000k, 90 CRI, 2200 LUX] ($3.50)
American Fluorescent Shop Lamp ($34)
Baby Plate ($4 used)
4 x #8 Screws ($?)

Total about $45 + TAX

The tubes I ultimately settled on where the GE Chroma 50 T12 48". They are inexpensive at only $3.50 each and available at Lowes and other hardware stores. Their CCI is 5000k with a CRI of 90 outputting 2200 LUX at stock power. The lamp housings I used were American Fluorescent electronic ballast shop lamps. They were the more expensive housings as far as shop lamps go, but I found them more attractive and ruggedly built than the really cheap lamps. It seemed like the extra few bucks were worth it. To attach the lamps to the C-Stand I screwed a "Baby Plate" to the top of each with short #8 sheet metal screws.

Recipe #2 ingredients:
4 x GE Chroma 50 T12 48" [5000k, 90 CRI, 2200 LUX] ($3.50)
Coroplast 1cm thickness
Aquarium VHO 4 lamp Electronic Ballast Kit + "German End Caps" (lamp plugs) And wiring ($160)
Velcro
Baby Plate ($4 used)
Clear Plastic Packing tape
Reflective Mylar Space Blanket ( $1.99)
3M "Super 77" Spray Glue

This recipe can be completed more economically by using multiple standard 4 lamp ballasts and wiring them to only 2 or even 1 lamp each. (Achieving an over-drive effect) Suitable electronic ballasts can be purchased for about 10 bucks each. The advantage of this scheme is that it allows running more or less lamps as required... A sort of poor-mans dimming. The "German End Caps" and handy because of the way they grip the tube, simplifying the mounting to the Coroplast with Velcro. Velcro is used to make tube removal easy. It's possible to remove the tubes and use them without the body/reflector.

Recipe 3:

Of these three recipes, the Compact Fluorescent is by far the easiest and most exciting. Since CFLs are self ballasted and all use electronic ballasts, 95% of you work is done by simply selecting the "bulb". CFLs also have incredibly high output for their size which is good if you are trying to get a "harder" light source. They'll never look like the sun or a Fresnel lamp, but you can at least get SOME shadows with a CFL in Par reflector. The only down side is that the best CFLs available seem to have a max of 85 CRI. It isn't a terrible CRI... but it is the minimum you should consider for film and video.

Research Continues:
Worth further investigation are the newer more efficient T8 and T5 tubes. In the end I didn't go with these because I was not able to find a local outlet for tubes of the correct specifications for a reasonable price. I went with T12s because they are common and inexpensive.

I've only just scratched the surface with over driving. It is an area that shows much promise.

Judging by their output vs. size, and how hot they seem to get I assume most CFLs MUST be "over driven" compared to regular tubes. Many of the commercial kits seem to use CFLs. I haven't even gotten close to exhausting the possibilities with Compact Fluorescents. There is currently a staggering selection of CFLs available with the correct CCI and CRI. It is worth further investigation.

Canon Cinema

Everyone seems to be in love the new HD shooting Canon SLRs (5D, 7D and Rebel 2ti). I have a 7D myself, and it's the coolest toy I've had in ages.

But depending on the "mission" it's not always really the best choice of video camera. Other times, the cameras unique strengths make it and ideal choice. Here are some of the cameras strengths

1. Small and lightweight
2. Incredible choice of lenses
3. Excellent stills shooting capability

It also has some major shortcomings:

1. Aliasing due to binning or skipping photosites on read-out of the sensor
2. Rolling shutter
3. Highly compressed h.264, 4:2:0 sub-sampled video
4. Poor ergonomics for video shooting
5. Poor on-board sound, which essentially necessitates dual system operation for anything requiring decent sound.

Each shortcoming is minor in itself, but they do add up. Third party vendors like Zacuto and Redrock Micro sell accessories to address some of the ergonomic short comings of camera. Of course, once you pimp the camera out with all the accessories, it's no longer small and light. And the technical issues are still a problem. The accessories do nothing to solve the the aliasing or rolling shutter.


Still, the success of these cameras have gotten Canons attention and now Canon seem to be listening to the concerns of video shooters. Thier recent addition of 23.98fps as a feature on the 5D is evidence of that. If Canon were smart, they would really go after this market because they could OWN it in two years with just a few careful moves:

1. Make a large sensor video/cinema camera similar to the Red Scarlet that takes Canon EOS lenses
2. Improve the ergonomics of the Canon EOS lenses to make them more "cinema like" . This could easily be achieved with the current line of lenses with clever electronic control of the mechanism. Third parties are already doing it on the Red One. But Canon knows their lenses better than anyone, so have an advantaged position in really making it work well.
3. Don't cheap-out on the output options. Consider lightly compressed I-Frame only 10 bit 4:4:4 or RAW for video recording, or at the very least, allow uncompressed 4:4:4 video or packed RAW to be sent out the SDI of HDMI port. HDMI 1.3 supports some incredible data rates, so in some ways it's more versatile than SDI.

and last of all...

4. Keep the price of the system in the "prosumer" range.

That's it. The other details would work themselves out in the planning of the new system. But in my mind, the important thing is to keep it compatible with the EOS lenses to gain the economy of scale that provides. Perhaps they could also create a new line of "super premium" lenses that are above even the "L" series. They could be "super speeds" marked in T-Stops and have improved cinema style ergonomics., etc. But I think they should still have electronic focus features, and be provided in EOS mounts so still shooters could still use them if they like. (That would help keep the volume up so they are still commodity lenses, like the L series)

gigabit NAS from a Walmart eMachine

One of my big projects/hobbies at the moment is building a digital archive of my all my families photos and videos. We are also digitizing all of my late father's artwork we can track down. (My dad was a very prolific artist and writer... we still haven't even cataloged his entire body of work!) The scans of all the old pre-digital slides, negatives and prints are being done by Scan Cafe. So far I've gotten on batch of scans back from them and I'm very happy with the results.

For the "I.T." side of my project, I'm using a multi-tiered system to maintain the stability of the data:
1. Master server with RAID-6 storage (also runs Digital Asset Management System)
2. Back up server with simple rsync mirror of the master server
3. Off-site Internet based "Cloud" backups.
4. DVD backups of high-value metadata and configs

The Backup Server does double duty and acts as the "Cloud Pusher" server.

At present I'm building the Backup Server / Cloud Pusher. Because none of the inexpensive cloud back up providers (i.e. Mozy / Backblaze) support Linux yet, the Cloud Pusher needs to run Windows. Windows is not my favorite OS. However, in this case it's a necessary evil.

The Cloud Pusher will be running 24/7, so low power utilization is also a requirement.

To address all the requirements I finally settled on using a refurbished Walmart special, the eMachines EL1300G-01w slim desktop. Compared to good contemporary hardware, it's a complete embarrassment...totally low-spec and slow. But for my purposes it is nearly perfect: It's inexpensive, comes with Windows pre-installed, and is fairly power efficient due to it's CPU and other components. (AMD Athlon 2650e / nVidia GeForce 6150se / nForce 430 / 2Gigs RAM ) Seriously, for what it cost, including the Windows License, it's a very good deal. ($180) I would not have been able to even build a functional computer sans OS for what this refurbished eMachine cost. Mass market hardware is so cheap these days!

However, the included 160 GB hard drive was totally inadequate for this project. I replaced it with a 2TB Hitachi.

To migrate to the new 2 TB drive, but keep the OEM install of Windows Vista, I used a fantastic live Linux distro called Parted Magic. Parted Magic is a Swiss Army Knife for disk partitioning, cloning, backup and system rescue. It also has a ton of tools for checking hard drives, system hardware, and doing other useful stuff. To be honest, due to recent versions graphical improvements, it's even become my favorite "Micro-Distro" / "Thumb Drive Distro". Parted Magic has everything you need to boot a machine and get a web browser running. ( When not at work, that's all I typically need anyway.)

Before I even got started with the migration to the new drive, the first thing I did was go though the OEM Windows install and remove the Bloat-Ware and "Craplets" that every OEM seems hell bent on pre-installing. Vista is notorious for it's sluggishness, but with tweaks it can be pretty usable. I went though and optimized performance by turning off un-used services, flashy graphics, etc. Honestly, those two exercises were the biggest use of my time. (And one of the reasons why I so detest Windows). I could have easily set up Linux in a FRACTION of the time required to whip this OEM Windows install into shape. Walmart and eMachines put so much crap on the machine is makes it almost un-usable. They should be ashamed. The Windows license was "free" with the computer, so I guess I had to "pay" for it somehow. But I can see why people have such low opinions of these cheap computers. Windows and the junk the OEMS install on them make the computers seem like complete junk. (When in truth, the hardware is actually pretty powerful)

Once the machine was all tuned up I was ready to start the migration. This is where things got a bit tricky. The EL1300G only has two SATA ports and both are used. (One for the DVD-RW and the other for the hard drive) I COULD have used a USB bridge board to hook the new hard drive up to the computer, but then performance would suffer. I would have also lost access to a lot of the low level control of the drive. What I did was unplug the DVD-RW and use the SATA port it was taking to plug in the new drive. Here is a photo of the computer with the case open and both hard drives plugged in:

[PHOTO]

To boot Parted Magic I used a USB Thumb Drive rather than a CD. At this point I have to comment on how awesome bootable thumb drives are. Not only are they relatively fast compared to an optical drive, they are so convenient! In this case, I could copy stuff from hard drive to hard drive over SATA, for best performance, even though I only had two SATA ports. (Meaning I could not boot from a device on either SATA port) I used Clonezilla (in Parted Magic) to clone the old hard drive from the old 160 gig to a the new 2 TB drive.

After the cloning was done, I re-partitioned the drive using GParted to re-claim the area that was occupied by the OEM "restore" partition, and reserved some new space for my own "restore" partition, which in this case is an disk image I created with PartImage. My new restore partition is also a reserved space for a potential future Linux install. Basically, the entire new setup has a little bit of "future proofing" built-in to protect for the case of a Linux dual boot setup. I haven't bothered doing the install yet, since this machine doesn't need Linux yet. But all the partitions required are in place.

So I guess the gist of this post: Love-fest for Cheap Hardware. Love-fest for Parted Magic. Check out the tool. It's super useful and a great example of how Free Software is completely superior to commercial software. Here is my finished NAS / Backup Server / Cloud Pusher

Long live Duchess.

The gardeners came this morning and when I went to check the coop on my way out to work the chunks of skin, feathers and the head were gone form the lawn. I'm guessing they mowed over the feathers and they got sucked up with the grass. They either mowed over the head and it was sucked up also or they cleared it. Anyway, there is much less evidence of what happened as of this morning. I'm glad I did the rounds last night because it would have been harder to figure out what happened otherwise. All that was left today was few feathers.

My concern now is that whatever hit Duchess now knows a good place to get chicken easily. It will be back at some point for seconds. Even if I DO lock the chickens up at night I'm going to have to make sure the coop is predator proof because whatever it was that got them is probably going to be more persistent next time. (Especially if it really likes chicken)

The sad life of a prey animal... At most it's only been a few days since she died and already there is almost no evidence that she ever even existed. Well, I mean, I have the chicks at least, so she lives on though them. But her body is pretty much totally gone.

I'll be much more careful with her offspring.

Duchess is dead

I was away for four days at a family reunion. A friend of mine was taking care of my chickens but since she could only come by once a day she couldn't open and close my chicken coop (The "Chicken Breasts 2000") I decided it would be OK to just leave the coop open the whole time I was gone. I usually close it up at night to keep the chickens safe from nocturnal predators. Chickens are very vulnerable at night because they really don't put up much of a fight when they are sleepy. It wasn't the smartest move, but I felt it would be better than leaving them cooped up, especially considering how hot it is right now.

Enough back story... I got back home today to discover my loss. After settling in I headed back to the CB-2000 and check on the hens and lock up the coop. Duchess was nowhere to be found. But I noticed some feathers scattered about. I widened my circle to include the lawn in the back yard only to find more feathers, chunks of flesh attached to feathers, and a chicken head. The chicken head was pretty beat up, so it's not like I can give a positive "ID", but I'm pretty certain it is (was?) Duchess.

Right now I'm thinking it was either a skunk or a raccoon that did her in. But you never know, a dog might have gotten in there during the day and gotten her. That's never happened before so I'm really leaning toward skunk or raccoon... something nocturnal. Considering that she seems to have been consumed almost entirely, I really doubt it was a domestic animal. ( Who knows, it might have even been a hard up chupacabra)

I'm pretty bummed about this. I would have been a little upset if one of the new hens was the one that got it but for Duchess to be the one to buy the farm is a real upset. IMHO she really deserved to live after giving me so many chicks and putting up with the two roosters for so long. The thing that really has me scratching my head is why that skunk or raccoon didn't ever eat any of the eggs in the past. Duchess was laying eggs in the bushes at one point for nearly a month before I noticed and as far as I could tell no pests ever bothered to eat any. Not even rats.

It was stupid of me to leave the hens unlocked at night. I probably should have moved them into the garage while I was away. Now I know there is SOMETHING that will attack and kill them at night so I will be more careful in the future. I just wish it didn't have to be duchess that was the first to go.

R.I.P. Duchess You will be missed.

Film Making

Lately I've been involved in a group at work that makes movies. We make little movies as practice for making big movies.

My little brother is also an aspiring film maker. He asekd me for some advice about doing a full feature. I thought the info might be interesting to others so I'm re-posting a more generic version of it here:

Even though I haven't ever done a REAL (i.e. money making ) feature project my self (Uh, outside of my job, but I'm just rank and file there) I have a few pieces of advice from my exposure so far.

There are two important factors: Story and Performance. Everything else serves those things, no matter how artistic it is ( i.e., lighting, cinematorgraphy, etc.) The story is what people walk away with but it's the performance that keeps them engaged moment to moment while they are watching the film. I'm sure you can imagine how a bad performance could kill the whole project.

Your are on the right track by wanting non-suck talent. The actors will bring a HUGE amount to the project. Be very picky about your casting and make sure you audition everyone. If they don't have a good resume and can't do the audition than forget it. You'll never get a good film out of them. (There are plenty of trained actors to work with so why work with amateurs?) Make sure that you are a non-suck director as well! You might want to take some acting workshops yourself, so you can learn how to spot "false" performances and get the best out of your actors. There are several books out there on directing. I'm reading one now called "Directing Actors" which is quite good.

In that same vein, you don't want your script to suck either. I'm not sure who's doing the writing for you, but make sure the script is solid before you start shooting. The right script with the right take can definitely turn heads. Napoleon Dynamite may not have made big bucks at the theater, but it definitely put those guys on the map. Much of that script seemed like it was simply accommodating the resources they had. But it did it in such a fantastic way! Don't forget the local resource you have when coming up with your script.

As for getting crew, I suggest you try to put non-students in the key positions on your crew. Get as experienced a DP/Cinematographer as possible. Same for the sound guy. Everyone else on the crew can be more green. They are basically just moving stuff around after all. But they'll still be learning by watching the pros work, and you film will be much better for it!

From a technical stand point, sound quality is more important that picture quality. Give sound proper attention. Bad sound will drive the audience crazy while poor picture will only bother them a little. "Bad" picture can sometimes even be justified as a stylistic choice... for example, skip bleach, cross processing, or shooting Super 8 will definitely have a detrimental effect on the image "quality" but given the correct stylistic intent the loss of "quality" might actually be desirable! Bad sound is just, bad sound. Make sure you have a good sound recorder and mikes with someone who knows what they are doing running them.

You can't over-prepare. Scout out locations and have a solid game plan at the very least. I personally would board everything out. I'd probably even do a 3D animatic to figure out what shots you can actually get. (Depending on the lenses you have, It might be physically impossible to place the camera for certain shots in a small room) If you have all your shots planned out in advance, you also wont have to shoot too much "coverage", which will save a bunch of time and tape/film. (The side effect is that it might make editing a lot easier as well) If you want to do any dolly shots, I would definitely do an animatic because they can be a huge time suck. Maybe you don't need that technocrane shot? The animatic will tell you in advance for a lot cheaper! The animatic could be just you working. Once you have your cinematographer, you can bring him in on it also. If you get a cinematographer who says he doesn't like to plan like that I wouldn't trust him. He should know there is room for improvisation and improvement on the set and should feel comfortable with planning. But the idea here is the get a really solid basic plan down so you can be honest about your schedule and get the best stuff on the day. When you are actually shooting, you, as the director, should be totally absorbed in getting the best performance from the actors. You don't want to have to worry too much about shot planning etc.


When you are working cheap like this the main problem is getting the commitments you need from the various cast and crew members to "go the distance". That's another reason why planning is so important. You need to be able to tell everyone exactly what the schedule is and how long a commitment you need from them. The shorter the commitment, the better the odds of getting a seasoned cast/crew. If the script is solid enough you might be amazed at the talent you can attract... assuming the commitment isn't too long.

Put a chicken to sleep

I know of a method for "hypnotising" chickens. You do it like this:


But this trick for "putting a chicken to sleep was new to me: