Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

"simple and ridiculous" dragon vs Mx test..

Status
Not open for further replies.
Gavin Greenwalt said:
Great simple test. One request if someone hasn't asked already would be an ETR test. Instead of using the extra resolution in the highlights I would be curious to see a shot where both peak the same and then show off the shadows. (Also 3:1 compression so that we can see the real noise not a de-noised image). It seems like most of the tests are matching the gray point but it would be interesting to see both ends by themselves.
Click to expand...

Yes, good idea. Now there will be folks who will take the shots in this thread and play with the ASA slider to show how different the sensors are. This does not unveil much because the sensors received the same amount of light and the sensors response is not altered.

Shoot the test again (no flare please) and use NDs to bring down the exposure step by step. I think in shots like these with this high contrast 3 stops difference Nd 3 - to ND9 will already tell a lot.

Thanks again for the shots, valuable and selling Dragon well.

Hans
 
Robert Ruffo New said:
What are your thumbnails exactly of? What do you prefer in Epic? Your post is not clear.
Click to expand...
the thumbs are simple crops of image areas...i prefer the epic image....did not print these but on the screen at 100% 16bit tiff epic looks better to me....or at least the noise in the dragon file is a turn off for me....yes it is at 2000 but the underexposed epic file looks cleaner to me...and that at 10:1.....afaik the promise was dragon is much cleaner (2000 is the new 800) and has much better compression....
what i am seeing in these tests is the there is at least one stop, probably/maybe 2 stops extra DR....noise, resolution, seems to be pretty similar.....maybe i am blind or stupid or both....
what it comes down to is probably this....the first scene (backlit) is something i do all the time....but i would not shoot it without fill...dragon or epic....the second scene i pretty much prefer the epic file as it is....
 
luigivaltulini said:
Paul, I'm sorry in what sense?
Epic Dragon is better to 80o iso, be cleaner.
Why prefer Epic?
Click to expand...
you are probably right, dragon at 800 should be as clean as epic at 800....but somehow i was under the impression that we were getting epic 800 noise "performance" (which IMO isn't that great anyway) on dragon with 2000....i just don't see that
 
luigivaltulini said:
Paul, I'm sorry in what sense?
Epic Dragon is better to 80o iso, be cleaner.
Why prefer Epic?
Click to expand...
luigi....i really want to thank you again for taking the time and doing and posting these....they are a great help...i am looking forward to seeing more....i will definitely download and play but probably won't post my opinion anymore, i really appreciate the honest discussion here and don't want to upset people or come across as purely negative....everybody has their own opinion and these things can be very subjective....
 
Paul...

I think that if you have a "good" (not pressed) Epic immage at 800, Dragon can handle that 1,25 stops below.

The examples I have seen so far is more like 6 stops below, which is not equal...
 
I think a fairer comparison would be a Scarlet Dragon to Epic MX because they are both 5K acquisition. Either that, or pop off some Epic Dragon at 5K.

Nobody that currently owns one is championing Dragon as a low light beast. It must be a push pull deal (smaller photons for more pixel/resolution vs. larger photons to capture more light), and right now RED is pushing 6K and dynamic range.
 
Hans von Sonntag said:
Yes, good idea. Now there will be folks who will take the shots in this thread and play with the ASA slider to show how different the sensors are. This does not unveil much because the sensors received the same amount of light and the sensors response is not altered.

Shoot the test again (no flare please) and use NDs to bring down the exposure step by step. I think in shots like these with this high contrast 3 stops difference Nd 3 - to ND9 will already tell a lot.

Thanks again for the shots, valuable and selling Dragon well.

Hans
Click to expand...

It might be better to simply stop down or change the shutter angle to decrease the exposure rather than using NDs. NDs are typically not precisely one stop or two stops and often introduce "SOME" color bias. It would be great to eliminate this as a variable in a test like this.
 
Eric Haase said:
It might be better to simply stop down or change the shutter angle to decrease the exposure rather than using NDs. NDs are typically not precisely one stop or two stops and often introduce "SOME" color bias. It would be great to eliminate this as a variable in a test like this.
Click to expand...

Changing the shutter manipulates the sensor. Probably negligible. My heritage is film... we could change the shutter from 180 to 172,8 degrees and with fancy cameras sometimes down to 45 degrees. However, 180 degrees were the golden standard. Hence the NDs. But you are right NDs colour the picture to some extend.


Hans
 
paul schefz said:
luigi....i really want to thank you again for taking the time and doing and posting these....they are a great help...i am looking forward to seeing more....i will definitely download and play but probably won't post my opinion anymore, i really appreciate the honest discussion here and don't want to upset people or come across as purely negative....everybody has their own opinion and these things can be very subjective....
Click to expand...

Paul, no.
Read the comments are valuable for anyone here in the forum.
The nice thing about this one Redusesr be, a community of sharing.
I see what you mean, but trust me that these two shots to be the extreme.
I hope you stay a comment.
 
paul schefz said:
what i am seeing in these tests is the there is at least one stop, probably/maybe 2 stops extra DR....noise, resolution, seems to be pretty similar..
Click to expand...

In my Log comparison (#98) I see approx. 2 stops in the whites and 1.5 to 2 stops in the blacks more DR in the Dragon picture. The flare in the MX shot does make the comparison somewhat unequal, however the results speak for them selfs.

In my experience the Epic MX sensor has a DR of approx. 12 stops. The Dragon in this shots exhibits the above mentioned 3-4 stops more DR which leads to the conclusion that it has a DR of usable 15-16 stops. This is a lot and a huge improvement.

Resolution is higher, no questions here. But will people use the higher resolution? Yes, sure. 5K-HD hass a significant oversampling advantage over the widespread used Epic MX 4K-HD regarding noise. However, in terms of sensitivity and noise Dragon seems to be pretty similar to MX.

This is all no wonder. Dragon has smaller pixels than MX due to the increased resolution. All things equal smaller pixel mean less DR and less sensitivity. This is the reason why Arri did not go the 4k route, they felt that DPs prefer DR over resolution. Considering that Dragon delivers significant more resolution AND DR the missing gain in sensitivity is bearable.

I haven't tested it yet but I guess that Dragon will cope with underexposure 1 Stop better than MX. I use the MX sensor in my daily routine with 400 ASA and in high contrast scenes like a sunny park with 800 ASA. I guess the Dragon can be exposed in such situations down to 1200 ASA delivering much wider DR than MX. However, the lowlight performance won't be much better than MX. Candel-light dinner scenes etc... work best with 320ASA on MX. Perhaps Dragon does this with 500ASA. All the numbers I mention here come from my experience to shoot noise-free pictures. People who don't hesitate to use de-noising in post my have different numbers, YMMV.

As a side note: In most tests I see graded pictures. Log pictures are much better for comparison. Grading always kills information, log not.


Hans
 
Hans...I think your interpolations are about spot on.

There is a bit more technically in the bottom, but it mostly adds richness (not noise) to the lower exposures.

WB in Dragon also works a lot better.

I am looking forward to Dragon optimized Color Science. I see there is a potential there which might extend useable DR downwards a bit. But that depends on what exactly the color science is. If it is only an adjusted gamma, it will probably not alter much. As "we all" start from log or linear.
If it bites into the calibration, it could change more things.

But I guess we will just hafta wait and see.
 
Hans von Sonntag said:
In my Log comparison (#98) I see approx. 2 stops in the whites and 1.5 to 2 stops in the blacks more DR in the Dragon picture. The flare in the MX shot does make the comparison somewhat unequal, however the results speak for them selfs.

In my experience the Epic MX sensor has a DR of approx. 12 stops. The Dragon in this shots exhibits the above mentioned 3-4 stops more DR which leads to the conclusion that it has a DR of usable 15-16 stops. This is a lot and a huge improvement.

Resolution is higher, no questions here. But will people use the higher resolution? Yes, sure. 5K-HD hass a significant oversampling advantage over the widespread used Epic MX 4K-HD regarding noise. However, in terms of sensitivity and noise Dragon seems to be pretty similar to MX.

This is all no wonder. Dragon has smaller pixels than MX due to the increased resolution. All things equal smaller pixel mean less DR and less sensitivity. This is the reason why Arri did not go the 4k route, they felt that DPs prefer DR over resolution. Considering that Dragon delivers significant more resolution AND DR the missing gain in sensitivity is bearable.

I haven't tested it yet but I guess that Dragon will cope with underexposure 1 Stop better than MX. I use the MX sensor in my daily routine with 400 ASA and in high contrast scenes like a sunny park with 800 ASA. I guess the Dragon can be exposed in such situations down to 1200 ASA delivering much wider DR than MX. However, the lowlight performance won't be much better than MX. Candel-light dinner scenes etc... work best with 320ASA on MX. Perhaps Dragon does this with 500ASA. All the numbers I mention here come from my experience to shoot noise-free pictures. People who don't hesitate to use de-noising in post my have different numbers, YMMV.

As a side note: In most tests I see graded pictures. Log pictures are much better for comparison. Grading always kills information, log not.


Hans
Click to expand...


I like what you said and I agree in full.
Arri knows that go up to 4k resolution involves considerable light on the problems of pixels.
In fact, I'm curious to see the next ARRI 4k.
I do not think there will be many new features, the difference in these levels is difficult and requires years of work ....
 
Scott C said:
Forgive me if i'm missing something dumb, but are the dots I'm seeing in the dragon image on the purple/blue squares hot pixels? If so I assume a blackshade would correct for that?
Click to expand...

was a speck of dust that was passing at that moment .. incredible ...
 
Of course the Dragon should keep more dynamic range and better color. But Im not sure why the Epic looks so hazy? Is it the olpf that is not as good, leading to light bouncing around more?
 
Status
Not open for further replies.
Back
Top