ARRI ALEXA LF - 4k, FullFrame announced

[Björn Benckert]

Don't talk bullshit. My goodness, I've heard very stupid things by professionals, and I'm just lost for words.

Do you even own a Red? Have you ever used an Arri camera? Arri has more DR than Red. Not even Monstro surpasses Alev3 in DR.

Have you personally done any testings to back up what you just wrote? I think before you use words as bullshit go do some testings as. I really would like to see any type of footage that shows that Alev3 has a wider range than monstro.

 

[Björn Benckert]

True and not true from what I understand and tested myself (I'm not a tech master so I don't have the math on it, but what I've experienced while grading both), the Alexa and Red has similar DR, it's just that the internal DR handling is different on the Alexa and it handles the rolloff into shadows better (rolloff into highlights works the same in IPP2) while rolloff into highlights works in conjuncture with how it process colors over the DR range. Arri points out that it has 14+ stops, which is a description of what I assume to be the workable DR while the extra stops makes up the rolloff handling. The Monstro says 17+ stops, but I believe it to be meassured to all DR stops regardless of rolloff.
With some clever tweaking in IPP2 you might come close to the Alexa DR style, but there are so many parameters to make that happen that it's almost impossible to just create a conversion LUT that makes Red look like Alexa. This needs to be done more internally, like with the DXL and Light Irons colors science.

What is PV doing internally? I really doubt they do anything else but swapping OLPF and using their own version of IPP2 luts. To me thats not "internally" and if you like it, then it very much shows that no its not about the camera, but more about how it´s used.

 

[Hrvoje Simic]

A qualification is required here.

Prior to Monstro
+) Arri's excellent signal to noise ratio (out of the box) had meant the usable dynamic range was felt to surpass RED's.
+) RED's sensor may however been able to record (buried in noise) a higher absolute dynamic range than the Arri.

Red Monstro
+) We have not seen a new DR Test for the Monstro.
+) We have not seen how effective the new Signal-to-Noise smoothing algorithm that Arri is use for the LF. With a greater resolution (ie 'fatter') negative - they can trade absolute image resolution for signal to noise - increasing the effective usable dynamic range available (out of the box).

AJ

1) The image is typically not made of black and white squares and typical DR tests are insufficient to show the camera performance.

2) The thickness of the negative is not defined by just signal to noise ratio nor spatial resolution, what is missing is saturation to signal intensity performance.

3) Spatial resolution oversampling and minimizing noise visibility that way affects the shadow part of DR. Not the highlight part. And as you decrease the pixel size full DR can be similar due to noise size reduction but the DR distribution cannot and mid point moves to the right, trading highlight range for shadow range.

4) Red cameras offer excellent imaging performance and great cinematic imagery if one learns to use them properly and adapt exposure and light to their DR distribution.

 

[Karim D. Ghantous]

Do you even own a Red? Have you ever used an Arri camera? Arri has more DR than Red. Not even Monstro surpasses Alev3 in DR.

Have I used them? Irrelevant - they have been tested side by side. The only thing that the Alexa has, as a package, is a very tightly managed output. RED's output is as raw as you can get, save for wavelet compression. It's the only 'digital negative' that's worthy of the term.

RED knew that MX, as great as it was, didn't have the edge over ALEV 3 in terms of outright DR, and they never shied away from that. Dragon either matched or beat ALEV 3, while Monstro is better again.

BTW I don't talk bullshit. I discovered long ago that there is no profit in being wrong or misinformed.

 

[Patrick Tresch]

1)
4) Red cameras offer excellent imaging performance and great cinematic imagery if one learns to use them properly and adapt exposure and light to their DR distribution.

The most problem I see is the colorist that use RED footage like Arri ones but it's another way to work with. There are much more creative and technical choices that you have to make an know with RED. But again it's not Rocket science.
Even more if he calls himself a professional.

Pat

 

[Aris_Gavriilidis]

I sense Red has something up its sleeve. Especially with their Gemini sensor and its unique abilities and also given that they've been able to produce two different sensors after Dragon in a small period of time that managed to perform better with smaller pixels, which in theory shouldn't be possible. They definitely have the capability to increase the headroom above middle grey. I'm willing to bet that if not with Monstro already, they'll definitely have something else to compete with a Mini version of the new LF camera. I've got faith in Red, don't just write them off in the presence of a new camera, no matter how disruptive it may be.

 

[Christoffer Glans]

No need in chasing Alexa look for Red to look fantastic and on the same level.

Alexa also looses its potential in standard route, btw.

Tell that to all the high profile names in the industry, including Roger Deakins. For most cinematographers, Red doesn't give the look they want, it doesn't seem able to, or else it would have been done already. Fact is still that the film industry is flooded with Alexas, there's a reason for that and it's not found in tech specs or pixel peeping comparisons.

 

[Brian Boyer]

3) Spatial resolution oversampling and minimizing noise visibility that way affects the shadow part of DR. Not the highlight part. And as you decrease the pixel size full DR can be similar due to noise size reduction but the DR distribution cannot and mid point moves to the right, trading highlight range for shadow range.

This is what I've been trying to stress, albeit a bit clumsily at times. Arri is NOT ready to shrink pixels because they know how important highlight latitude is to the look and success of the Alexa. They're not going to give that up.

It's the one thing no one has matched or surpassed them on and it's a major part of their bread and butter appeal, along with color.

4) Red cameras offer excellent imaging performance and great cinematic imagery if one learns to use them properly and adapt exposure and light to their DR distribution.

The words "learns" and "properly" are key to understanding RED's challenges. People are lazy at times. When things don't work as expected or desired, people tend to move on to the things that do.

Many people are willing to learn but that doesn't necessarily mean they're learning how to do a thing properly or in a way that yields the most benefit.

Speaking of a willingness to learn, we often comment on how many features and episodic TV shows RED loses to Alexa but maybe worse than that are the shows that start out shooting RED for Season 1 and switch to Alexa for Season 2. The worst part is the shows look noticeably better. Sorry, but they do.

That's not to say it's definitive proof that Alexa looks better than RED, only that crews often know how to get better looking results out of Alexa than they do RED.

I know Blue Bloods switched. Although, to be fair, they switched from R1MX to Alexa, which was no contest. Wynona Earp switched. Famous In Love recently switched. I saw the teaser promo for Season 2 and right away I noticed how much better it looked than Season 1.

Knowing better than to assume the crew upped their RED game, a little sleuthing uncovered a BTS photo on their Twitter page that shows they made the switch. You'll know the one I'm talking about.

In my experience, there's not that much to learn when it comes to using RED cameras. Other people find it more involved, I guess. From my rather extensive observations, I'd say RED cameras have historically been less forgiving than Alexa when it comes to certain aspects of the image.

They've been less forgiving with certain types of lighting and lighting conditions, with certain colors, and particularly less forgiving in terms of makeup. That last one is huge because it greatly impacts the perception of skin tones, which is something else I've been saying for a while that gets ignored.

Take The Good Doctor on ABC. That show has some horrendous makeup. It's particularly bad on Hill Harper's character, but everyone, especially the people of color, have hit-and-(mostly) miss makeup jobs. The only BTS footage I could find showed a RED camera on Steadicam.

I don't know if they shoot the entire show on RED or just the traveling shots, but if it's all RED, that's a problem. Despite shows that successfully shoot on RED and look fantastic the vast majority of the time, this show may blame the camera like so many other shows tend to do and switch to Alexa for Season 2.

And there will be a Season 2 of that show.

 

[Hrvoje Simic]

Tell that to all the high profile names in the industry, including Roger Deakins. For most cinematographers, Red doesn't give the look they want, it doesn't seem able to, or else it would have been done already.
.

It may not seem but it IS able to.

It has not been done before in the scope necessary to allow that impression to flourish. It has been done recently. Actually finished few days ago. Things take time.

I'll show you in a different thread.

Fact is still that the film industry is flooded with Alexas, there's a reason for that and it's not found in tech specs or pixel peeping comparisons.

That is true and the main reason for it is the fundamental difference in philosophy. Arri has always been primarily focused on film philosophy. Red aims for versatility and multifunctionality. Main vision behind Red cameras was making a high-end compact multifunctional camera. And the vision was realized. There is nothing that versatile with that level of motion imaging performance in that compact body.

This pre-determines functionality, practicality and type of workflow. And some types of shoots have specific requirements. Where no one wants or has time to fiddle with menus or black shading for example and introduce more riskful variables in scenarios where delays and mistakes have accumulative effect and a high price. Smaller productions and single shooters have many luxuries which larger productions do not.

Plus there is brand perception of course and 100 years of cinema history has some weight here as well.

 

[Milano Questa]

Have I used them? Irrelevant - they have been tested side by side. The only thing that the Alexa has, as a package, is a very tightly managed output. RED's output is as raw as you can get, save for wavelet compression. It's the only 'digital negative' that's worthy of the term

RED knew that MX, as great as it was, didn't have the edge over ALEV 3 in terms of outright DR, and they never shied away from that. Dragon either matched or beat ALEV 3, while Monstro is better again.

BTW I don't talk bullshit. I discovered long ago that there is no profit in being wrong or misinformed.

http://www.arri.com/camera/alexa/workflow/working_with_arriraw/arriraw/format/

 

[Sabovic Adis]

Tell that to all the high profile names in the industry, including Roger Deakins. For most cinematographers, Red doesn't give the look they want, it doesn't seem able to, or else it would have been done already. Fact is still that the film industry is flooded with Alexas, there's a reason for that and it's not found in tech specs or pixel peeping comparisons.

AFAIK, deakinses don't do color work, high or not, and are not the ones that are able to satisfy (or not) their needs or the needs of somebody else by pushing the right buttons. You know, there are colorists...and then there are colorists. And that's a fact, it's true.
Roger that? :biggrin:

 

[Antony Newman]

If a LARGE pixel on a sensor could 'count' 1000 photons each frame before clipping, was compared to another sensor where 4 SMALL pixels each with the same 1000 photo counter (and where the smaller pixels were 1/4 the area of the large pixels):

4 x SMALL pixels would be able to count 0-4000 photons per frame in the same area that the 1 x LARGE pixel could count 0-1000 pixels.

However as wonderful as it is to have 4 x the Dynamic Range (in this example) - there is huge drawback : You had to deal with 4 x as many pixels; 4 x as much accumulation, 4 x the bandwidth, and a lot more HEAT that you would have to dispense with, in addition to noise and power.

My speculation is that Arri would have reduced the size of their pixels (which might even IMPROVE their dynamic range capabilities), but have not yet resolved the Wattage requirements in the pixels processing.

AJ

Edit : More specifically : If Arri could create a 6k Sensor with dimensions as their 4K LF sensor for a UHD output, and in the same power budget -> I believe they would.

<snip> Arri is NOT ready to shrink pixels because they know how important highlight latitude is to the look and success of the Alexa.

 

[Aris_Gavriilidis]

Can someone here please grab a Xyla chart and a Monstro? Would make things (a bit) easier. I definitely would if I had either of those things at hand!

 

[Eric Lange]

Can someone here please grab a Xyla chart and a Monstro? Would make things (a bit) easier. I definitely would if I had either of those things at hand!

+1


(Got nothing else).

 

[Björn Benckert]

Can someone here please grab a Xyla chart and a Monstro? Would make things (a bit) easier. I definitely would if I had either of those things at hand!

The thing is it does not help much if you got one of those two things, you need to have both at once.

 

[Eric Lange]

The thing is it does not help much if you got one of those two things, you need to have both at once.

Ha ha ha ha ha (That's hilarious).

I think the problem is they ARE mutually exclusive. Once you buy a well kited out Monstro/Monstro VV probably the last thing you want to point your camera at IS such a chart... (Whistling, hands in pockets, looking sheepish); There's a whole universe out there to image/shoot other than a Xyla chart.

It's those that have not invested / ponied up yet that DO want to see such charts lol.

Rental is different environment entirely.

 

[Björn Benckert]

This is what I've been trying to stress, albeit a bit clumsily at times. Arri is NOT ready to shrink pixels because they know how important highlight latitude is to the look and success of the Alexa. They're not going to give that up.

Ok explain: "highlight latitude"

What do you mean with that? Are you referring to the numbers of stops you use between your highlights and the sensors clipping point. Arri does not set that margin, the user does.

Also clipping point and pixels size has little or nothing to do with each other. If you look at an early digital camera that had few and large pixels, those got flooded quite quickly. But things moved and are moving forward, so today we got not only pixels that are more light sensitive and has greater range then before... The photo sites also come smaller and in more dense configurations. The ALEV sensor is old and it not so dense.

My guess, This LF sensor is Arris last squeeze of the ALEV line of sensors and very soon they will announce something with higher density. And then all that large pixel talk will be as forgotten as the explaining tha HD is enough and or looks better than 4k.

 

[Hrvoje Simic]

Also clipping point and pixels size has little or nothing to do with each other.

You are mistaken.

If you look at an early digital camera that had few and large pixels, those got flooded quite quickly.

Yes that was in the infancy of digital.
And today Alev3 holds most highlights and also has the largest pixels by a strange coincidence.

But things moved and are moving forward, so today we got not only pixels that are more light sensitive and has greater range then before...

Light sensitivity is not the only aspect of DR.
Ability to hold charge before overflooding is another. And that determines the upper side of DR.

The photo sites also come smaller and in more dense configurations.

That is not a DR advantage.

 

[Tom Gleeson]

Can someone here please grab a Xyla chart and a Monstro? Would make things (a bit) easier. I definitely would if I had either of those things at hand!

Thats only the start as you really need to set up a rigorous, accurate and detailed test procedure to be of any value. Stand alone testing only tells you so much and IMHO comparative tests are more valuable. I would prefer a test comparing cameras in identical situations. Alexa as its industry gold standard against a Monstro and the Venice maybe? I would expect Monstro to pip the Alexa at the post for DR although the Alexa LF may grab the crown? Rather than charts I would also prefer a real world situation with under and over exposures on a face. I have considered doing this test but DR is only one measure of camera performance and all these top end cameras are capturing never before imagined dynamic range so arguing over who has a half stop advantage when we are in the 15 stop range is a bit "whatever."

 

[Mike P.]

My guess, This LF sensor is Arris last squeeze of the ALEV line of sensors and very soon they will announce something with higher density. And then all that large pixel talk will be as forgotten as the explaining tha HD is enough and or looks better than 4k.

But that's kinda what he's saying; Arri are *not* going to release a smaller-pixel/higher resolution sensor that performs *only* as well as the ALEV3, but at 8k. They're likely spending more time tweaking/improving their next sensor so it's more than simply an "8k s35 ALEV3" (especially if they want it to last 10 years as well)... And the ALEV3 being top dog for 10 years has allowed them the freedom to take their time with the next sensor (whereas RED have had, what? 4 different sensors in that same time period, none of which have supplanted the ALEV3 in terms of colour or effective/useful DR.)

I think anyone saying 'but Dragon/Helium/Monstro is just as capable as ALEV3, so you don't need it to look like Alexa' seems to be avoiding the one of the main reasons Arri and the ALEV3 have done so well. The best example I can give is the DXL; it uses the same RED hardware but Panavision felt it necessary to re-developed it to literally act more like an Alexa (or 'act more like film' if that's an easier pill for RED diehards to swallow.)

From panavision, "Light Iron Color 2 deviates from traditional digital color matrices by following in the footsteps of film stock philosophy instead of direct replication of how colors look in nature"... That's very similar to what Arri have done with Alexa/ALEVIII from the start, and it's worked incredibly well for their products even though they apparently have inferior specs on paper... I'm inclined to think there is no reason RED couldn't have developed an official "RC5219/RG5219" so that the footage didn't look "mathematically better" than it's competitors, but so that it felt and behaved more like a reference film stock in the same way Arri did with the ALEV3 and the same way Panavison is doing with the DXL.

Again, IPP2 is a step in that direction anyway, but I still think they're focusing more on maximizing a mathematically correct image (and hoping everyone is going to get their picture professionally finished), rather than making the best image possible out of the gate (only going back to the all-encompassing RWG/L3G10 footage to re-develop it another way if a "5219" reference image doesn't suit their needs.)

"Just" is a term which tends to roll over the tongue easily to oversimplify the unknown.
There is no "just" in signal shaping and colour science. It would be great if it was that easy.

It's unwise to assume "simple" means "easy". And if RED sensors/cameras capture more DR/colour/resolution than ALEVIII (as RED has claimed since the MX), then it is "just" a (presumably not easy) matter of making the cameras react/respond in a more film-friendly manner. This is especially true when trying to pull users from the ALEXA camp. Again, the DXL is doing just that and the response from the industry and cinematographers seems to be universally positive.

"Full blown Alexa mode" = unrealistic

Okay, so replace "Full blown Alexa mode" with "full blown filmstock mode", as it seems being 'more like Arri' is what bothers you. Back to my hypothesis, if RED released an official "RC5219/RG5219", I'm sure even diehard Alexa users would take a peek and want to test it out. And if it actually had a film-like colour/response as good as Alexa's (e.g. -7/+7, good brightness/saturation, film colour instead of "real colour"), it'd definitely win a few hearts and minds (because of form-factor, resolution, weight, ease of raw, file sizes, and all the other objectively positive things RED have going on).