Disclaimers first: if this has been covered to this level of detail, please point me to the appropriate pages because I’ve waded through hundreds of posts but couldn’t find specifically this. Also, this is the result of a week of abusing Google with absolutely zero real world experience.
I’m pondering two things with Scarlet. Circles of Confusion (Blur - DOF) and Low Light performance. My aim is classic film style shorts and features in which the subject is often isolated. No run and gun.
COC
Below follows a little write up of the analysis:
1. Firstly I confirm the numbers against known SCARLET footage; then
2. I manipulate it to see what is needed to replicate typical feature shots on 35mm by using 2/3” chip. I didn’t want to assume that opening wider and shooting longer solves it all because it might mean that one must open up to an f stop of 0.8, shooting long and thus pushing the talent further away. I think it’s obvious that there could be some complexities in achieving that.
1. FIRST THINGS FIRST:
I would like to confirm my number crunching. I have all the formulas for the Circles of confusion (Coc) and DOF calculations. I’ve confirmed them with the charts too. I’ve looked at the original izzy shots of two years ago and this is what I’ve got: A 100 mm prime was set at T5.6. Izzy was 4-5 ft away. The tail seems to be another 3 feet further.
I’m switching to metrics now. Using subject at 1.3m and tail at 2.3m:
- Theoretical Coc calc gets me to = 0.647mm;
- The Coc that our human eye is comfortable at accepting to be in focus for 35mm is around 0.03mm (considering how far the typical picture is blown up). Four thirds at half that and Two thirds at another half which is 0.0075mm. This means that the Coc at Izzy’s tale is many times that which is acceptable, meaning we think it’s really out of focus; and
- A Coc of 0.03 leads to a DOF of 5.7 cm. Considering the size of Izzy’s head and the Coc that grows on her right side, I’d say the numbers look right.
2. NOW MOVING ON
I doubt that the sensor would change because the impact on just the fixed lens and the primes is going to be significant. Thus, assuming fixed lens Scarlet to be released having a focal distance of 28-224 but limited to T2.6 on widest with a 2/3” chip.
Because I am use to full frame, let’s say I’d like to frame a shot with a subject at 2m, an object behind at 3m from the lens. Thus 1m apart. A 85mm lens at 5.6. The Coc at the 3m object is 0.224mm. That means we are out of focus but less than the Izzy example. The acceptable DOF here is 0.19m.
To replicate that shot on the fixed Scarlet I need to cut the focal by 4 due to the crop to have the same frame or FOV or composition. So that is 21.3mm. I’ll open up to 2.6 because fixed can’t go further. BE AWARE that hyperfocal is at 5.7m so we are dangerously close to having an enormous DOF. The subject and object remain at the same distance. Coc is 0.029mm at the furthest object. That is very disappointing because it is not remotely as blurry. The DOF 1.6m. The other way would be to us the prime to get to T1.8 (the 16mm or the 25mm because there is no 21mm planned.) In that scenario I’d rather push to the 25mm. The Coc would be 0.0586 which is double of the fixed but a lot less than on full frame.
It is obvious that just opening wider and shooting longer is not as simple as said. You would need to really reconsider how you frame or compose shots. You want to shoot longer focal but that chops the FOV. Also, we all know the benefit of primes, but considering the Coc problem it has a distinct advantage, hence the additional need for an interchangeable Scarlet.
NB – the hyperfocal issue was more significant than what I initially thought because the shorter the focal the closer it gets, which means everything is in focus. An example is: if the focal is 16mm and the T2.6, the hyperfocal is at 3.3m. For 25mm it’s at 8m. An obvious advantage for run and gun but also a significant disadvantage for isolating subjects.
LOW LIGHT
I’m used to full frame. We use the Nikon D700. It has a 12.1 MP chip which is 4256 x 2832. A pixel is thus around 8.45microns (square). The Scarlet is 5MP at 3072 x 1620 which leads to a pixel which is about 2-3 times smaller. I have no idea what the signal to noise ratio is like on the Scarlet, but I know the D700 is amazing and that we can easily push to an ISO of 1600.
Does anybody have an idea what this will be like? Any numbers I can work with?
SUMMARY
If this is correct, which I am not sure of, then it leaves me with no other choice:
RED, PLEASE can you put a bigger chip on the Scarlet, or create some sort of very low entry Epic for the nobodies who would love to shoot features on tiny budgets? VERY BIG PLEASE?
I’m pondering two things with Scarlet. Circles of Confusion (Blur - DOF) and Low Light performance. My aim is classic film style shorts and features in which the subject is often isolated. No run and gun.
COC
Below follows a little write up of the analysis:
1. Firstly I confirm the numbers against known SCARLET footage; then
2. I manipulate it to see what is needed to replicate typical feature shots on 35mm by using 2/3” chip. I didn’t want to assume that opening wider and shooting longer solves it all because it might mean that one must open up to an f stop of 0.8, shooting long and thus pushing the talent further away. I think it’s obvious that there could be some complexities in achieving that.
1. FIRST THINGS FIRST:
I would like to confirm my number crunching. I have all the formulas for the Circles of confusion (Coc) and DOF calculations. I’ve confirmed them with the charts too. I’ve looked at the original izzy shots of two years ago and this is what I’ve got: A 100 mm prime was set at T5.6. Izzy was 4-5 ft away. The tail seems to be another 3 feet further.
I’m switching to metrics now. Using subject at 1.3m and tail at 2.3m:
- Theoretical Coc calc gets me to = 0.647mm;
- The Coc that our human eye is comfortable at accepting to be in focus for 35mm is around 0.03mm (considering how far the typical picture is blown up). Four thirds at half that and Two thirds at another half which is 0.0075mm. This means that the Coc at Izzy’s tale is many times that which is acceptable, meaning we think it’s really out of focus; and
- A Coc of 0.03 leads to a DOF of 5.7 cm. Considering the size of Izzy’s head and the Coc that grows on her right side, I’d say the numbers look right.
2. NOW MOVING ON
I doubt that the sensor would change because the impact on just the fixed lens and the primes is going to be significant. Thus, assuming fixed lens Scarlet to be released having a focal distance of 28-224 but limited to T2.6 on widest with a 2/3” chip.
Because I am use to full frame, let’s say I’d like to frame a shot with a subject at 2m, an object behind at 3m from the lens. Thus 1m apart. A 85mm lens at 5.6. The Coc at the 3m object is 0.224mm. That means we are out of focus but less than the Izzy example. The acceptable DOF here is 0.19m.
To replicate that shot on the fixed Scarlet I need to cut the focal by 4 due to the crop to have the same frame or FOV or composition. So that is 21.3mm. I’ll open up to 2.6 because fixed can’t go further. BE AWARE that hyperfocal is at 5.7m so we are dangerously close to having an enormous DOF. The subject and object remain at the same distance. Coc is 0.029mm at the furthest object. That is very disappointing because it is not remotely as blurry. The DOF 1.6m. The other way would be to us the prime to get to T1.8 (the 16mm or the 25mm because there is no 21mm planned.) In that scenario I’d rather push to the 25mm. The Coc would be 0.0586 which is double of the fixed but a lot less than on full frame.
It is obvious that just opening wider and shooting longer is not as simple as said. You would need to really reconsider how you frame or compose shots. You want to shoot longer focal but that chops the FOV. Also, we all know the benefit of primes, but considering the Coc problem it has a distinct advantage, hence the additional need for an interchangeable Scarlet.
NB – the hyperfocal issue was more significant than what I initially thought because the shorter the focal the closer it gets, which means everything is in focus. An example is: if the focal is 16mm and the T2.6, the hyperfocal is at 3.3m. For 25mm it’s at 8m. An obvious advantage for run and gun but also a significant disadvantage for isolating subjects.
LOW LIGHT
I’m used to full frame. We use the Nikon D700. It has a 12.1 MP chip which is 4256 x 2832. A pixel is thus around 8.45microns (square). The Scarlet is 5MP at 3072 x 1620 which leads to a pixel which is about 2-3 times smaller. I have no idea what the signal to noise ratio is like on the Scarlet, but I know the D700 is amazing and that we can easily push to an ISO of 1600.
Does anybody have an idea what this will be like? Any numbers I can work with?
SUMMARY
If this is correct, which I am not sure of, then it leaves me with no other choice:
RED, PLEASE can you put a bigger chip on the Scarlet, or create some sort of very low entry Epic for the nobodies who would love to shoot features on tiny budgets? VERY BIG PLEASE?