House pisses me off

[Nils Ruinet]

Enabling the Highlight Tone Priority & winding the contrast all the way down increased the DR of the Canon by 1.5 stops

That's interesting. I'll have a look at this next time I have to shoot with one. Do you know what Gamma curve they used ? Was it with a standard one or a Super-flat / Hypergamma-like curve ?

One issue I noticed while testing was that if you use a too flat curve, you get a decently low-contrast image, but while grading (and of course adding contrast again), all those nasty compression artifacts became really apparent (plus loads of banding in some situations). And I don't see how to avoid that...

 

[Nils Ruinet]

As the noise floor increases, you loose usable DR. There is a reason why Kodak & Fuji make films with different ISO ratings!
I think you are confusing a feature with a benefit!

No, I was just reacting to the fact that you said the DR changes when you change the ISO on the Red, which is not true. Nor does the usable dynamic range. I'm not saying it's good or bad, it's just a fact.

Of course you're going to be digging more in the shadows at ISO 2000 than at ISO 320, so you'll get more noise, no surprise, but that has nothing to do with DR, which is the number of stops you have between the noise floor and the highlights clipping. When chosing a higher ISO, you get less usable latitude in the shadows, but equally more in the highlights, so DR stays exactly the same. That was my point.
I know you know that. :smile:

 

[Deanan]

Thats exactly the problem with R1, every stop you increase the ISO is a stop of DR lost, a Canon DSLR does it differently!

On the Canon DSLRs, if you increase ISO a stop you do lose a stop of DR. On RED you don't. The DR stays the same regardless of ISO.

 

[M Most]

Fine. But my clients don't know this either. One of them told me he was sure that 24 was shot 100% on the 5D. You see what I mean? The net has really taken off with misinformation about this and the clients are not DOPs, nor is it their job, but they like to feel they know something, and the net makes them feel that they do. BTW - kudos to Canon on their social networking-based marketing coup.

The Internet is a source of widespread misinformation about everything. 24 was shot on the 5D (wrong). Film is dead (wrong). Videotape is dead (wrong). Red has taken over network television production (wrong). Canon DSLR images are awful, unwatchable, and cannot be used for anything, anywhere (wrong).

You can't stop people from reading and being gullible enough to accept as fact the nonsense that's on the Net. Getting angry about it is a bit of a waste of energy.

 

[M Most]

No, you clearly don't know me at all. I didn't say 5D footage looks good blown up or that the 5D compares favorably to higher-end cameras on overall image quality. I know that it doesn't, that's why I own a Red.

Well, I will say it. I worked on an Internet distributed project called The Ballad of Mary and Ernie, directed by Robert Stadd (a visual effects supervisor who has done most of Michael Mann's pictures over the last few years). It was shot on the 5D and was projected on a reasonably large screen at Laser Pacific. I had color corrected it on a video monitor (it went to SR tape) and thought it looked pretty good, but had some of the aliasing issues that have been mentioned here. When I saw it projected on a 2K projector, I was somewhat shocked that not only did the image quality hold up, it looked better than it did on the monitor, and most of the "issues" were less noticeable, not more noticeable. Now, I wouldn't recommend using these devices for high end work for a multitude of reasons. But making blanket statements that are not borne out by the actual experiences of those who have used them is to be just as closed minded as those you're criticizing.

There's theory, FUD, and reality. I prefer the latter.

 

[Stephen Williams]

On the Canon DSLRs, if you increase ISO a stop you do lose a stop of DR. On RED you don't. The DR stays the same regardless of ISO.

Are you telling me that if I can get +/- 5 stops of latitude at 160 ISO & I will also get +/- 5 stops of latitude at 1280 ISO?

EDIT My testing implys that if I have a scene of 10 stops of dynamic range, I will get a different result if I exposed it at 160 ISO or 1280 ISO, If the dynamic range remains the same at higher ISO's I would get the same results just with more or less noise.

 

[Stephen Williams]

That's interesting. I'll have a look at this next time I have to shoot with one. Do you know what Gamma curve they used ? Was it with a standard one or a Super-flat / Hypergamma-like curve ?

One issue I noticed while testing was that if you use a too flat curve, you get a decently low-contrast image, but while grading (and of course adding contrast again), all those nasty compression artifacts became really apparent (plus loads of banding in some situations). And I don't see how to avoid that...

I think it was the standard curve, we did very little final grading, I created the look with the lighting.

 

[CJ Roy]

Are you telling me that if I can get +/- 5 stops of latitude at 160 ISO & I will also get +/- 5 stops of latitude at 1280 ISO?

Technically yes, since the ISO is meta data and not baked in. However, dependent on how processed, you should lose those stops depending on how far you're pushing it. Especially on the M sensor.

 

[Kemalettin Sert]

EX1's are only approved for SD production.
5D's Not approved for anything
Red is approved for use by Independent productions only

http://www.bbc.co.uk/commissioning/production/hd.shtml

i recently watched documentary on BBC shot on Canon HDSLR..

 

[M Most]

Very ironic and elitist thread.

Rewind two years and you will find the exact same thing with RED v Film.

People were pissed that their $xxxxxx cameras were losing rental/production value to a camera that had problems/workaround (eg, offered less DR, operational problems, bad workflow etc etc.... yes, we know those issues are generally solved now).

Canon are doing what RED did.... they have just done it in a sub $4k market, rather than a sub $40k market.... that brings even more severe problems/workarounds.

About a year ago, I said something that is still the case:

It's fun to be the disruptor. It's not fun to be the disrupted.

 

[Subhadip Sen]

i recently watched documentary on BBC shot on Canon HDSLR..

I remember seeing a BBC test a while back where the 5D failed the very first test and no further tests were performed on it.

Even the Canon HF10 made it through that for subsequent tests.

 

[Kemalettin Sert]

http://www.dvxuser.com/V6/showthread.php?t=202932
you right they dont accept these cams for HD production

 

[Nils Ruinet]

Are you telling me that if I can get +/- 5 stops of latitude at 160 ISO & I will also get +/- 5 stops of latitude at 1280 ISO?

No, not exactly.
If you have +/- 5 stops of latitude at 160 ISO, that means you have a 10 stop DR, right ?
Well at 1280 ISO (+ 3 stops), you'll get -2 and +8 stops latitude, which still equals to 10 stops of DR.
Less latitude in the shadows, more in the highlights, same DR.

 

[Jim Retar]

Jim, this is a professional forum where we discuses technical aspects of digital filmmaking within a wide range of topics, mainly related to the Red cameras (that is why is called Reduser). I think is not polite call “elitist” to a group of professionals trying to understand why a high budget production will switch from film to a low quality digital format. That is the kind of sentences that move the moderators to close treads like this one.

I was talking about the initial post.... not the replies or the people discussing this.... just the initial subject. Maybe I should have worded it better....

btw - it is not elitist to ask the question you posed, but that was not the original point of the thread.... it is elitist to think Canon are ruining the market and no one should use these cameras for broadcast work... especially before they have even seen the results.

 

[Stephen Williams]

No, not exactly.
If you have +/- 5 stops of latitude at 160 ISO, that means you have a 10 stop DR, right ?
Well at 1280 ISO (+ 3 stops), you'll get -2 and +8 stops latitude, which still equals to 10 stops latitude.
Less latitude in the shadows, more in the highlights, same DR.

So at 1280 ISO I have lost 3 stops of usable DR......Latitude is both over & under to be useful in my experience.
The Canon makes use of analog gain........... The director I was working with noticed that & chose the Canon.

 

[M Most]

Are you telling me that if I can get +/- 5 stops of latitude at 160 ISO & I will also get +/- 5 stops of latitude at 1280 ISO?

That's not what Deanan is telling you. He's telling you that the ISO setting on a Red changes where the midrange grey value is when the image is processed - in other words, it's only a metadata setting. The point on the sensor where the noise floor becomes unacceptable and the point at which it clips highlights - in other words, the actual hardware - is always the same regardless of the ISO setting. That's what defines dynamic range. Latitude is based on where you put your midrange grey, which is dependent on the ISO setting. If you increase the ISO, you've got less range in overexposure but more range in underexposure. If you decrease the ISO, the result is the opposite. Cameras like the Canons adjust ISO by changing the bias, which changes the actual characteristics of the sensor, which changes the actual dynamic range. There are some in-between ISO settings that are simulated by a combination of metadata application in post processing and actually manipulating the sensor characteristics, but for the most part, it's the latter. When you change those characteristics, the dynamic range changes.

At least, that's what I THINK Deanan was trying to say..

 

[Stephen Williams]

That's not what Deanan is telling you. He's telling you that the ISO setting on a Red changes where the midrange grey value is when the image is processed - in other words, it's only a metadata setting. The point on the sensor where the noise floor becomes unacceptable and the point at which it clips highlights - in other words, the actual hardware - is always the same regardless of the ISO setting. That's what defines dynamic range. Latitude is based on where you put your midrange grey, which is dependent on the ISO setting. If you increase the ISO, you've got less range in overexposure but more range in underexposure. If you decrease the ISO, the result is the opposite. Cameras like the Canons adjust ISO by changing the bias, which changes the actual characteristics of the sensor, which changes the actual dynamic range. There are some in-between ISO settings that are simulated by a combination of metadata application in post processing and actually manipulating the sensor characteristics, but for the most part, it's the latter. When you change those characteristics, the dynamic range changes.

At least, that's what I THINK Deanan was trying to say..

Hi Mike,

I am on the same page as you, not sure what Deanan was saying.

Stephen

 

[Nils Ruinet]

So at 1280 ISO I have lost 3 stops of usable DR......Latitude is both over & under to be useful in my experience.
The Canon makes use of analog gain........... The director I was working with noticed that & chose the Canon.

I think you're confusing latitude and dynamic range... Your usable dynamic range stays at 10 stops (in your example) whatever ISO you chose.

The analog gain method used by Canon has downsides too : it means your highlights will probably be clipping much earlier when set to 1280 ISO instead of 160 ISO, so you're actually throwing away highlight information.
While with the Red you would keep all the highlight information you had at 160 ISO. Pretty important too in my opinion.

 

[Stephen Williams]

I think you're confusing latitude and dynamic range... Your usable dynamic range stays at 10 stops (in your example) whatever ISO you chose.
.

Hi,

I always assumed that 'dynamic' had some connection with 'dynamics', which is why I assume the +/- dynamics should be connected to ISO.

Latitude used to the term that described the +/- error in an exposure that could give an acceptable result, now it's often (usuallly) used to describe DR.

 

[Brooks]

I was excited by the House announcement, mostly to see what they can do with the 5DmarkII. I've got a 7D, and I absolutely love it. But I consider myself a student filmmaker, and want to showcase my stuff on the web. And because the only money I have for a production is whatever I can save up, maybe a few thousand at a time, the video DSLRs are great for me. I've been writing scripts for years. But very few people will read a script these days. So it's better for me to produce a short based on the script, and if done well, use that to generate interest in the script and raise money to make the feature. The disadvantages of DSLR are minimized because I'm putting my stuff on the web.

But when the time comes to have a real budget, and shoot a feature, there's no way I'm going to shoot on a DSLR. Went to the RED event last week, and saw the res chart they put up on the big screen. The 5DmarkII footage breaks down really fast at like 1.3K. The codec is a mess, the aliasing, moire, etc. It's just not viable for anything that needs to be screened theatrically. And even at 1080p, the problems are visible.

So for those of us that need to practice the craft, and get good-looking footage for an ultra-low cost, then right now the DSLRs are a great tool. I think where people are getting upset is when the producers are suggesting them for larger budget productions and those who should know better (DPs) are going along with it. It's one thing to do experimental shorts, but it's dangerous I think to run a professional production with these cameras at the core.

Got to play with the Epic a little the other day, damn... 5K in that hand-held rig. It's going to change everything. Can't wait.