11 March 2007

Letus HD100 and the RED radio TVC

It's almost certain that for directors and cinematographers that learned the filmaking process during the celluloid dynasty era, the usage of shallow depth of field became consciously or subconsciously part of their cinematographic vocabulary. I am now, totally converted to a digital believer but it's difficult sometimes not to feel that the toolset is incomplete. At least for people that can't afford a Panavision Genesis® or an Arriflex D20.

For the rest of us, a DOF adapter will do just fine!

When i searched around for an adapter that would work with the JVC HD-100E, i wanted one that would look straight to the CCD. I didn't want to mount it on the Fujinon stock lens because of it's chromatic aberrations and the eventual length of the setup. I also wanted to avoid messing with reversed images as it would make the procedure unnecessarily complicated for the way i work. The only adapter that covered all that having a very affordable price was the Letus HD-100.

Ordering the contraption was very easy. Quyen Le at www.adapterplace.com always answered my emails immediately and the adapter arrived only 4 days after i ordered it. This was a big surprise considering that it had to be assembled and travel from USA to Greece! It arrived in a nice cardboard box filled with cushioning material. There were no instructions inside but Quyen was always available to answer questions and the adapter was very easy to use.

The lens mount was set for Canon FD because at the time i only had 2 Canon lenses to use with it. An 1.8 55mm and an 1.8 24mm. The adapter's mount can change from Canon FD to Nikon AI in less than 2 minutes by unscrewing a bolt and flipping the mount ring.

My DP and I went to test the Letus by shooting exteriors in Athens during a warm and clear afternoon sky. The moment we saw in the viewfinder what the HD100 was recording through the 55mm we knew that this thing really worked!

The first impression of the image's quality was very positive. No flicker. No visible vignetting and no distortion. Also, it was almost completely silent! The only funny thing we noticed was that when the adapter's iris was 100% open it had the tendency to bloom the highlights. By turning it 1/2 stop down the effect disappeared. It didn't concern us... It's worth mentioning at this point that if the iris is left 100% open for a few seconds it will slowly push itself down a 3/4 stop!

By reviewing the footage later at home i noticed that there was something like a very faint static grain visible on some shots. We thought it was some sort of dirt that we missed to clean. It was also obvious that the 24mm Canon could not resolve enough resolution for 720p (because of the 1.9x magnification?) but the footage looked good downscaled to PAL SD. That, gave me the confidence to try the adapter in real work situations so when the RED RADIO commercial came in, there was no question about using it.

Although the commercial had a budget lower than 25.000$, we tried to make the most we could out of it. Pre-planning is always valuable and saves time. I designed the storyboard until the shot count came down to 19 which was a number we could deliver in the available 12 hours. 1 hour more would mean the overtimes would drive the production overbudget.

The camera set up was simple. The footage was recorded to my 80GB DR-HD100 Firestore without using a tape in the camera. My DP was nervous about not using tapes for backup but for psychological reasons difficult to explain i just don't like using tapes in my HD100. I have done many TVCs with the Firestore and never had the slightest problem so call me crazy but i didn't worry too much about it...

We fed a 9" Sony BVM-D9H5ME HD monitor through the component output so that the clients could see what we were shooting. The 4:2:2 non-mpeg2 output was fantastic! I wish i had a way to use that for the editing! I have to find an affordable solution for doing that soon...

I usually operate the camera myself. I used the focus assist mode for focusing but the DP was always double checking the shot on the HD monitor in case i messed up. The Focus-Assist is not assisting very much in dark situations and the particular shoot was very dark. We didn't use any rods support system for the Letus because the Canons where not heavy and the adapter is very steady. The camera assistant was pulling the focus effortlessly and the Letus never joggled.

I don't find the light loss from the Letus a big problem. Actually, i welcome it for doing day exteriors but the boxing ring location was not very bright. We didn't really have a problem but ended up shooting most of it with the iris almost wide open!

Most of the handheld shots where done with the Fujinon lens. We also used the WCV-82SC wide converter for one shot. If we exclude the 24mm, the rest of the lenses mixed well together in editing. It's barely visible in the 576p version but in the 720p it's obvious that the 24mm Canon was the weak link.

My biggest fear during the shooting was that fast action in dark environment combined with the static noise (that didn't go away after cleaning the lenses) would drive the MPEG2 encoder crazy. Miraculously there wasn't any shots that suffered heavily from it but while i was grading the final cut i noticed some frames during very fast actions that made me wish i had the 4:2:2 footage instead.

The static noise was not very visible before the color correction. After adjusting some color curves the "dirt" would become a little more noticeable. Some of it had to be removed from two shots and it was a little tricky but i managed to get rid of it using Fusion.

After the shooting i tried to solve the static grain issue. At that moment i was sure that there was something wrong with the lenses so i decided to change them. I bought a Nikon 55mm and a Sigma 24mm (with Nikon mount) and to my surprise there was a huge difference in resolution compared with the Canon glass. The Nikon resolved as much resolution as the Fujinon with minimal C.A. and the 24mm Sigma was as good (or better) as the 55mm canon was. I am not saying that Canon glass is worthless. Maybe there was something wrong with those particular ones i had.

I would shoot some resolution charts (ironically made by Canon) to display the difference but the Canons are not under my possession any more.

Although Nikon made me very happy, the static grain did not go away. It was obvious now that it's source was the Letus adapter so something had to be done.

I contacted Quyen for advice and then i decided to open the adapter and clean its internal elements. I carefully cleaned as much of the mirrors (used to reverse the image) i could. Some of them were too deep inside to reach so i let them as they were. I assembled the adapter and tried it again. The static grain was reduced at least by 30% but it was still there.

Is it because i couldn't clean all the mirror elements? This is one of the two theories i have but i think the other one is the most possible. According to that the static "dirt" is coming from the moving ground glass. The ground glass' grain in my Letus is not very fine. Although it disappears when it vibrates i think that at some moment during it's vibration pattern it becomes visible. I will try to have the GG replaced with one of thinner grain and see what happens.

I found the Letus to be a great tool and i use it ever since with great results. Also, taking under consideration how much it costs, it makes me forgive the static noise pattern. It is not always visible but i hope a solution will be found for it.

The commercial can be downloaded from this place.

28 October 2006

Magic Bullet Deartifacting VS Chroma Blur

Discussing about 4:2:0 colour sampling recently, i realized that there is still the misconception that Magic Bullet's deartifacting tool is a fancy name for chroma blur. Well, it definitely isn't. It might be difficult to see the difference just by looking at a frame but there is a way to demonstrate very clearly what each tool does.

Here is a frame from a recent green screen work i did, shot with a GY-HD100. The reason i chose this footage is because it will make it easier to see the 4:2:0 colour compression.

Let's see the RED CHANNEL (magnified 4 times from original frame size). The colour artifacts are visible, especially along the hat's edge.

Now, let's see what the Deartifacting tool does...

The improvement is easily noticeable. The algorithm filled all the missing colour information and fixed HDV's compression artifacts.

Now, let's see what the Chroma Blur tool does. (The NLE used for both tools was VEGAS 7.0b)

It is clear that although the colour artifacts were smoothed out we created a new problem to the footage. We created colour bleeds and that means that we lost the footage's colour sharpness.

Compared with how bad 4:2:0 artifacts can be for your footage you might say that a little loss of colour sharpness is not such a big sacrifice, but if your aim is to produce the best possible image from HDV compressed footage then Magic Bullet's deartifacting tool is the better choice...