Alexa v Epic anamorphic resolution comparison Update

Arri have now released the Alexa Plus 4:3, bringing full frame Anamorphic capture with the ability to record Pro-Res to SxS cards (with an imminent firmware update) as well as Arri Raw (to a codex recorder). This is digital anamorphic shooting come of age; and the convenience of shooting to cards while capturing the full anamorphic image makes this the ultimate camera IMHO and outweighs any perceived advantage with the resolution of the Epic.

Arri had already released the Alexa Studio with a 4:3 sensor and a rotating mirror shutter giving the camera an optical viewfinder. As anyone who shoots with digital cameras knows, electronic viewfinders are a complete pain, and the idea of having an optical viewfinder back is very appealing; however having tested and worked with the camera I find the opposite to be true. The addition of the mirror shutter adds considerable weight to the front of the camera making it heavy and unbalanced. viewfinder flicker now seems annoying, being now used to seeing exactly what you are recording while you are shooting, it seems a step backwards to not having the actual image in-front of your eye, the camera can only display an image in the viewfinder or through a video out…not both. Arri have addressed the other problem of having a very dark image due to a stack of NDs in-front of the lens (common when shooting native 800asa) by building NDs into the camera behind the lens. There are rumours of a vastly improved EVF on the way, and one of these teamed with an Alexa Plus 4:3 has to be the way forward.

Alexa v Epic anamorphic resolution comparison

(NB: These videos are 1080p. They should be viewed with full screen selected on a monitor big enough to view full HD) 

This is a simple comparison of the final resolution of an image created with a 2 x anamorphic lens on an Arri Alexa and a Red Epic. The reason for this test is that the Epic captures a much higher resolution image than the Alexa, but how do they compare once they are both resized and rendered at 1080p, the usual delivery standard for something shot for TV. I hasten to add this is not a worthy test for something intended for projection.

The Epic 5K sensor captures an image size of  3226 x 2700 pixels. This is full height and cropped left and right to an academy ration as covered by the lens. The 16×9 chip does not cover the full height of the lens, but it does cover slightly more than the Alexa.

The Alexa 1920×1080 sensor (shooting to cards, not 2K Arri Raw) captures an image size of 1290 x 1080. The Alexa’s sensor is physically slightly smaller than the Epic’s, so to maintain the same angle of view as film the focal length of the lens would increase by a factor of 1.33 on the Alexa and 1.22 on the Epic.

All this equates to an image area 44% smaller than film on the Alexa and 33% smaller on the Epic. So on paper it seems to be all about the Epic…however that’s just numbers…

The Alexa image was captured to SxS card ProRes4444 in LogC at 25fps, then resized to 1920×1080 and processed in After Effect with a Log to Lin conversion and a tiny tweak to the colour balance.

The Epic image was captured in anamorphic mode with a 3:1 compression at 25fps, then resized to 1920×1080 and processed out through Red Cine X converting to REDcolor2 and REDgamma2 with a tiny tweak to the colour balance.

The same lens was used on both cameras. A Panavision 35mm USG at T5.6.

The most obvious thing is the size of the images. The Epic’s larger sensor gives a wider angle of view; a definite advantage. The difference in resolution is also apparent with the image from the Epic appearing slightly sharper than the Alexa, but only slightly; not the massive difference you might expect from the numbers.

So in conclusion the Epic does seems to win out but only just. If you prefer the Alexa as a camera then there is no reason not to shoot anamorphic on it, although the logistics of de-squeezing the image on set is much easier with the Epic. When Arri release the Alexa studio with it’s 4:3 full size chip covering 100% of a 2 x Anamorphic lens we should have the ultimate digital anamorphic camera.

UPDATE

I have since had a chance to look at these tests in a grading suite with a projector, and as far as resolution or the sharpness of the image is concerned there is very little to separate them. I should note that there seemed to be some color issues with the Alexa image, but it seems confined to this test and i have not seen this in any other Alexa footage. Arri have also update the Alexa’s firmware and it can now de-squeeze the image.

Thanks to Pat Wintersgill at Creativity Media.

Alexa Video

Epic Video

Thanks to Trevor Henen for his excellent assistance, and to Tony and Adam at Panavision UK.

Colour Experiments in Black and White

Inspired by old Technicolor 3-Strip colour film processes, an urge to process colour images at home and a free weekend; i decided to live it large and produce some colour images using only black and white film.

The principles are simple; a colour image can be photographed and re-produced by recording luminance values of the primary colours of visible light Red, Green, and Blue. This can be done by shooting 3 frames of the same scene onto black and white film, each one filtered to isolate Red, Green, or Blue; then scanning each frame and applying them as the appropriate colour channel in a RGB image in Photoshop.  I did this using three Lee polyester filters: 25 Tricolour Red, 47B Tricolour Blue and 58 Tricolour Green. Each of these filters completely isolates the single colour by filtering out all other wavelengths. I measured the transmission levels of each filter with a spot meter and used NDs to balance them, so they needed equal exposure compensation. I then mounted these filters in a row on some card to make them a quick as possible to change.

ColBW_01

ColBW_02

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I shot the first test on a Hasselblad 500 with a motor winder to make the time between exposures as short as possible; i used the 5×4 camera pictured later.

I chose Kodak T-Max 100 for it’s fine grain; the final image having three layers of the emulsion i wanted to minimise the grain. The film was processed normally in Agfa Rodinal.

RED FILTERED IMAGE

ColBW_03

 

 

 

 

 

 

 

 

 

 

 

 

GREEN FILTERED IMAGE

ColBW_04

 

 

 

 

 

 

 

 

 

 

 

 

BLUE FILTERED IMAGE

ColBW_05

 

 

 

 

 

 

 

 

 

 

 

 

After scanning the three images i created an RGB image in Photoshop and imported the three scans in as separate layers (make sure they are all separate layers and one is not the background layer). In the layer blending options (ctrl click the layer and select blending options), i assigned each image as the correct channel under advanced blending.

ColBW_06

 

 

 

 

 

 

 

 

 

It is important to keep track of which image is what colour throughout the process….and that is basically it; i now had a full colour image. A little moving around of the layers to get the channel alignment as good as possible (it will never be perfect due to the time between exposing each image) and lifting the levels a bit produced the image you see here.

ColBW_07

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Anamorphic distortion comparison

Hawk V-Lite 35mm and Panavision C-Series 35mm, barrel distortion comparison. Photographed at the same stop and distance. From the test we can see that the C-series lens suffers much less barel distortion than the hawk. This is in part due to the different designs of the lenses. This is explained in more detail in the next post.

Under projection the lenses are similar in contrast but the Hawk is sharper wide open; from T2.8 1/2 to T22 however the lenses are very similar. Both lenses are soft around the edges as you would expect from an Anamorphic lens this wide.

The Hawks are also very large, heavy and expensive. This 35mm is 5.3kg, and a full set of lenses is around half a million pounds! The C-series on the other hand are tiny in comparison, much more akin to the size of a spherical prime, but due to the fact they have not been manufactured for a long time they are irreplaceable so arguably priceless.

The other interesting thing to note about these two lenses are that the Hawks are currently manufactured and were designed only a few years ago, where as the C-series were designed and manufactured in the late 50s early 60s!

Anamorphic lens breathing and flair comparison

This is a side by side comparison of a 50mm Panavision C-Series Anamorphic lens and a Hawk V-Plus 50mm Anamorphic.

With the C-series the image stretches vertically only, as the lens focusses, where as the Hawk changes the shape of the image horizontally as well as vertically. This is mainly due to the completely different designs of the two lenses. The C-series is essentially a spherical prime at the back end and an Anamorphic front element; between the two however is an Astigmatic correction lens that moves on a gear; this corrects the lateral breathing as the lens focusses.

The Hawk on the other hand is again essentially a spherical prime at the back, in-front of that is the Anamorphic element (in the middle of the lens), then infront of that is a diopter as the front element that moves back and forth to focus. So the Hawk focusses without any correction resulting in lateral and vertical breathing.  The other obvious difference in these two designs is the way the lenses flair.

With lens flair it should be first noted that Anamorphic lenses are often chosen for the way they flair, despite lens designers trying to minimise it as much as possible. Films such as Close Encounters, Alien more recently Transformers deliberately use anamorphic lens flairs as a cinematic style to great effect (but that’s a topic for another time).

The flairs in these films are caused by intense sources of light hitting the convex anamorphic element of the lens. Due to the fact that the C-series have the anamorphic element on the front of the lens it is exposed to alot more of the light source thereby creating a much bigger flair across most of the image. The Hawks on the other hand have the anamorphic element in the centre of the lens shaded from being hit by as much of the light source giving much less flair.

Conventional wisdom then would indicate the Hawks are better at dealing with flair; and this is true, but ironically, and maybe due to the films I have mentioned, anamorphic flairs have become very popular and DPs want their anamorphic lenses to flair as much as possible.

So both lenses are very good; personally I find the lateral breathing of the Hawks an issue but some people will not. As for flairs you either love them or hate them…. I love em!

Just to be fair ‘The Lives Of Others’ was shot on Hawk anamorphics, a film I think is beautifully shot.

Film Stock Comparison

This video is 1080p at full screen if you have a big enough display (make sure you have HD selected); you really need to see it at this resolution to appreciate the differences properly.

Kodak 500T (5219), Fuji Eterna 500T (8573), Fuji Eterna 500T Vivid (8547), side by side comparison 4-perf Anamorphic. I exposed all of the stocks at 500 asa and they were processed normally.

I have always favored Kodak stocks because i consider them to have a finer and nicer grain structure and better colour rendition. It is clear however that these new stocks from Fuji have adressed these issues and i think there is little between these stocks now. The Fuji Vivid is a brand new stock that aims to produce very saturated colours and deep blacks; a bold move in the modern world of film scanning. Fuji demonstrated this stock to me recently with a theatre projection of a film print; the stock did indeed show very strong and well balanced colours with deep blacks and very little grain. What these results show from my point of view are that Kodak still seems to have a slightly finer grain and possibly a little more separation in the blacks whilst retaining very good colour rendition. The differences are however very slight and i am sure it would be very hard to tell them apart once they were graded properly.

The ultimate decision then comes down to price. Fuji have always sought to sell their stocks cheaper than Kodak, and producers inevitably favour Fuji for this reason. Now with such little difference between them the onus is on Kodak to drop the price as choosing Kodak is now very hard to justify as a DP.

4-Perf Anamorphic v 2-Perf Spherical comparison

This video is 1080p at full screen if you have a big enough display (make sure you have HD selected); you really need to see it at this resolution to appreciate the differences properly.

A side by side comaprison of a 4-perf Anamorphic standard 35mm negative against a 2-perf Spherical super 35mm negative. Explanations of the difference in resolution are explained in the previous post.

NB one can really appreciate the differences in a theatre projection; i’m afraid H264 doesn’t really do it justice.

Film Stock and Lens Test

 

I have been testing film-stocks and lenses for a future feature project.

For lenses I compared the Hawk V-Plus Anamorphics with the Panavision C-Series Anamorphics. Both are 2 x squeeze lenses, projecting a full height, squeezed image onto an Academy ratio standard 35mm negative; and un-squeezing horizontally by a factor of 2 to produce a 2.35:1 ratio image.

For stocks I compared Kodak Vision 3 500T (5219) with Fuji Eterna 500T (8573) and Fuji Eterna 500T Vivid (8547), shot 4-perf anamorphic and 2-perf spherical. These are the latest 35mm stocks from Kodak and Fuji.

StockAndLens

 

 

 

 

 

 

 

 

Reflected Exposure Values (500asa/24fps/180º)

Film Stock Comparison.

Thanks to Russel Kennedy for his expert assistance, and willingness to don that lovely necklace (to highlight chromatic aberrations…honest!)

 

HDR Experiments – conclusions

I would have liked to have had more contrast in the scenes we tested. The intention was always to try and eek out as much detail from the darkness to arrive at an interesting look. I feel this was achieved, and it is a look that could only come from two cameras (i have tried to match it with one video stream) and it has ‘hyper-real, high-res, extreme clarity’ HDR qualities.

So…was it worth it?

Well that depends upon your point of view. It is obviously a very expensive process. The 3D rig is expensive, you need twice the camera gear, it takes ages to setup, lens changes are hilarious, and you need alot of time in post. The camera set up is a beast and hard to work with; it is appropriate to think of the process as akin to 3D; it is hard to move the camera, so statics are good, and it makes sense to cut less like 3D. You can achieve a look, however than is unachievable in any other way, and for this reason, yes of course it is worth it!

I would like to do one more test in high contrast daylight, so watch this space.

HDR Experiments – 2nd test

The object of the second test was to A, create two perfectly synced and aligned video streams, and B experiment with HDR at night to see if enough detail could be eeked out of the shadows to obtain a pleasing image.

As mentioned before we used two Red cameras so they could be gen-locked and for the RAW files. One was a standard Red One and the other a MX Red, the MX set at various high ASA ratings to provide the shadow detail, and the standard camera left at it’s standard 320ASA and exposed for the highlights. We used a full size Element Technica 3D mirror rig. Once the rig was assembled and set up (a very long process) we had two perfectly aligned images. Rather than just point the cameras out of the window we decided to make something of it and put all the gear on the top deck of a Routmaster and drove around central London at night, shooting our subject lit and un-lit against the windows, looking into London’s city lights. Suffice to say getting the built mirror rig onto the top-deck of a bus was a giggle, but it all went remarkably smoothly. Results and conclusions to follow.

A huge thanks to Tony, Adam and Christian at Panavision UK for accommodating my ridiculous ideas; thanks to Alex Taylor for his faithful assistance, and to Simon Levene for bringing his directorial prowess to the project.

PanavTony1

PanavTony2

PanavTony3

HDR Experiments – MX Red Sensitivity

Testing the ultimate sensitivity of the Red MX camera. These stills are taken from some footage shot at 6500 asa 25fps/180º. The ‘dark hole’ (in Jeff Brown’s office) was barely enough light to get a meter reading. Suffice to say the next test is Red MX HDR; two Reds (maybe only one MX) on a full size 3D rig shooting some city street night exteriors. Thanks to Jeff Brown of Brownian Motion.

redNoise2 redNoise1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

HDR Rig 01

This is the rig we used to create the HDR footage in the first test. It’s a miniature 3D prototype rig (Element Technica) that is being developed for 3D on Steadicam. The unit is wonderfully small but seeing as Canon’s don’t work very well for HDR we’ll need a bigger one. We started testing with a 50/50 beam splitter mirror but trying to align the cameras is very difficult, and it would be a bitch to use on set, so with the plethora of 3D gear around at the moment the answer was simple.

HDRrig01_2 HDRrig01_1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

HDR Experiments – first test

As a first test and really a proof of concept, the above footage was produced using 2 Canon 5Ds perfectly aligned together on a 3D mirror rig, set at different exposures. The footage was then synced, split into two image sequences and then processed through Photomatix and re-conformed into a video sequence.

The Canon’s are not ideal for this however. They cannot be gen-locked together so there is some ghosting, and the files are heavily compressed. Next test will be Reds as they can be gen-locked and the RAW files will have a lot more information.

HDR Test Original Video Files from Ben Moulden on Vimeo.

Thanks to Adam and Tony at Panavision UK.