RE: Red Scarlet-X vs NEX-FS100?

[J Bellari]

I don't see any point in comparing a 4K camera to a 1080p body,
different audience and workflow.

Complaining that an F1 race car gets worse mileage and requires a pit crew,
doesn't make a Honda Civic neccessarily the better car for racing...

To be fair, when the new 4K monitors come out, we can put upconverted footage 1080p
from the FS100, F3 and Canon C300 up against native Red Scarlet 4K
(approx 3200 lines measured accoring to Graeme Nattress) and see how they compare.

Though ultimately pointless, would be fun...

 

[Ruben/Sheri]

Fs100 seems more cost effective who really needs 4k at this moment just our opinion

 

[S. Matthews]

different markets, what do you shoot? what do you want to shoot? what matters most to you?

both great camera's but both cater to two very different areas, I would love the fs100 for the low light but for me that is about it so it's not worth it, not saying it's other features are not appealing just not for what I'm looking for.

I much prefer raw video, and higher resolution (I'd much prefer to be oversampling than undersampling) but that is just me we all have our different styles.

4k after debayer is 3.2k measured resolution, but what 1080p cmos camera actually measures their full resolution, there is a difference between a resolution container size and measured resolution.

The fs100 and scarlet are both great and at different price points and feature sets, nothing wrong with that, but to say flat out one is better than the other for every person in the world is just wrong in my opinion.

have a good one!

 

[Ruben/Sheri]

mehh

 

[jetswing]

There's no point in comparing. Get both!

 

[Ruben/Sheri]

Just seems to us that RED would just be suited for narratives or something where the return of the project is very profitable, fs100 is perfect cost wise for independent internet distribution. I mean all that work/money talking almost 20k with lens and hard drive space etc.. on RED just to end up showing ur work compressed as hell on youtube or vimeo at 1080p, soo much money/work just for a few more pixels that most normal ppl wont even be able to tell the difference IMO, and the work flow is so slow, I have worked with R3d's getting ur footage off the ssd mag is so time consuming not to mention the cost of rigs you would have to buy being that RED is about 30lbs with lens and rig . yes the DR is amazing but unless your planning to do some outrageous grading not worth it in the end the, original concept scarlet seemed to be a better option but for the price and workflow and battery life you get from the fs100 man talk about s35mm bang for ur buck with an excellent picture quality and low light performance! So in conclusion if your gonna make one hell of a movie someone will watch in theateres or some studio will distribute get RED no question, but if all ur gonna do with it is shoot web stuff like scenery an weddings no point in getting RED.

 

[TimurCivan]

Scarlet runs about 21k all in.

It costs 4x as much. It performs about 4x as much. That's about right.

Redcode 6:1, 4k, 3k HDRX, 5k stills, built like a tank. Ur def getting what you pay for.

With regards to delivering 4K, you can deliver quadHD, right now on youtube. It looks pretty good, and can be played off existng internet infrastructure, on a moderatly fast computer. Meaning it can be embedded on your "film" page, website, industrial webpage, etc in scaling resolutions. When 4k panels arrive (very soon, there were several of them at NAB this year, and there are consumer 4k projectors coming in a few months) your content will be ready to go. 4k is a lot closer than you think.

 

[rx8pilot]

Scarlet is not and never will be 4k. It is a camera that measures well (3.2k by RED's hand), but IMO is about 2.5k in real world shooting. The FS100 pixel and masking arrangement is not publicly available I don't think but it is something higher than it's output of 1.9k. The frame size of any camera output has nothing to do with it's performance - including resolution. Technically Sony could output 8k frames from and FS100 and it would offer no improvements. In all likelihood, the Scarlet would spank the FS100 in most tests, but as many have pointed out it is about 4x the cost. It's just not 4x the camera. If you don't believe me, shoot a real world piece (not rez charts) with both cameras. Blow up the FS100 to 4k (properly) and finish them both in 4k. Display them on a nice 4k display to a group and see who notices a difference. If it is literally 4x the camera - my mom would notice a difference. The difference from SD to HD is 4x and it is striking to anyone with functioning eyeballs. In this case my guess is that you would have to look very close.

A few years back I was the DIT with an F35 on a huge camera test for the ASC and the PGA. We setup every imaginable test for many cameras over two days and then spent months in post sorting it all out. In the end, the "4k" Red One looked softer than the 1.9k F35. Why? I can't say for sure, but in the end we do not shoot charts and our eyes tell us if it is good or not.

Comparing output resolution of 4k to 1.9k by the numbers makes no sense because any image can be scaled to fit.

 

[Ryan Patrick O'Hara]

You also need to think about post-production if you're doing it. Can you handle 4k footage?

With the exception of transcoding speed, most people can work with Red footage just fine. Most editors, especially professional editors, don't edit in 4k. There is really no need to edit with anything greater than 1080 for the purposes of editing. Also deliverables are rarely 4k, so going back to 4k after the edit is rare and not really an issue if you kept good care of your project and file structure.

Do a one-light grade, transcode to ProRes codec, edit, source to original 4k files, color correct, and done. The actual editing/realtime playback process only required the handling of pro-res. Easy does it.

 

[Ruben/Sheri]

4k tv's, projectors are out, but will not be affordable to the general public anytime soon, i have to disagree the tech is too early and wont be affordable not anytime soon, sorry but most ppl still only watch dvd's and barley are buying there first hd tv and first bluray players, so you really think ppl are going to dump all there blurays and tv's for new 4k stuff sorry 4k has a long way to go in the consumer market, not gonna be affordable for a while mark my words 4k is far from being a regular thing, most tv shows still broadcast sd, so no I don't think 4k anything will come out at an affordable price anytime soon they will be too expensive and only for studio work and rich ppl most movie theaters only project no more than 2k fact! 4k has a while to go sorry but I think 1080p is fine for now anything else is over kill for the web, Fs100 is perfect for what it is, 4k is great for cinema work but lousy for youtube and Vimeo because of the compression rate not worth the money for just a few more mostly unnoticeable pixels for normal ppl witch cant really tell or don't care about 4k, look how long it took for 1080p and bluray to catch on! overkill in my opinion but that's just what we have seen and heard, search and tell me how much a 4k projector or monitor costs might as well take out a second mortgage on ur house, one day yes 4k will be more widely adopted but not for a while sorry $76,000 for a 4k monitor right now fact , 1080p is good enough for now and at some point the human eye wont detect higher rez at some point it wont matter the human eye can only process so much info....

Scarlet runs about 21k all in.

It costs 4x as much. It performs about 4x as much. That's about right.

Redcode 6:1, 4k, 3k HDRX, 5k stills, built like a tank. Ur def getting what you pay for.

With regards to delivering 4K, you can deliver quadHD, right now on youtube. It looks pretty good, and can be played off existng internet infrastructure, on a moderatly fast computer. Meaning it can be embedded on your "film" page, website, industrial webpage, etc in scaling resolutions. When 4k panels arrive (very soon, there were several of them at NAB this year, and there are consumer 4k projectors coming in a few months) your content will be ready to go. 4k is a lot closer than you think.

 

[mico]

Scarlet is not and never will be 4k. It is a camera that measures well (3.2k by RED's hand), but IMO is about 2.5k in real world shooting. The FS100 pixel and masking arrangement is not publicly available I don't think but it is something higher than it's output of 1.9k. The frame size of any camera output has nothing to do with it's performance - including resolution. Technically Sony could output 8k frames from and FS100 and it would offer no improvements. In all likelihood, the Scarlet would spank the FS100 in most tests, but as many have pointed out it is about 4x the cost. It's just not 4x the camera. If you don't believe me, shoot a real world piece (not rez charts) with both cameras. Blow up the FS100 to 4k (properly) and finish them both in 4k. Display them on a nice 4k display to a group and see who notices a difference. If it is literally 4x the camera - my mom would notice a difference. The difference from SD to HD is 4x and it is striking to anyone with functioning eyeballs. In this case my guess is that you would have to look very close.

A few years back I was the DIT with an F35 on a huge camera test for the ASC and the PGA. We setup every imaginable test for many cameras over two days and then spent months in post sorting it all out. In the end, the "4k" Red One looked softer than the 1.9k F35. Why? I can't say for sure, but in the end we do not shoot charts and our eyes tell us if it is good or not.

Comparing output resolution of 4k to 1.9k by the numbers makes no sense because any image can be scaled to fit.

Great post Carlos. would love to hear all your thoughts on the FS100 and how you came to them.

 

[mico]

With the exception of transcoding speed, most people can work with Red footage just fine.

Yeah but that transcoding is a mother and should be elaborated on. Just copying the files over for safety redundancy is brutal and then a transcode. If you got alot of footage, oh man time gets warped. 64GB gives you 23minutes of 6:1 compressed 4K 24p footage. And until theres an affordable computer that makes the red rocket card unnecessary I will never submit to that time suck. Jannard was asked many times for inclusion of a in cam 1080p option but his stubbornness for 4k all day every day makes one wish that he build and offer a time machine fast forward to the future module as an add on. But that would probably cost another $5K.

 

[Michael Olsen]

Define "real" resolution. The number of pixels on the sensor? In that case, the Scarlet is definitely a 4K camera, and the FS-100 is probably in excess of a 2.88K camera. The number of pixels in the output image? In that case, the Scarlet is definitely a 4K camera, and the FS-100 is a 1080p camera. Or the number of pixels resolved under a standardized set of circumstances? Any mosaic pattern single-sensor camera won't resolve the full resolution of it's sensor. So in that case, the Scarlet is (conservatively) a 2.5-3.2K camera and the FS-100 a 1.6K camera. Any way you want it, that's pretty apples to apples.

In all likelihood, the Scarlet would spank the FS100 in most tests, but as many have pointed out it is about 4x the cost. It's just not 4x the camera.

To be fair, however, the price/performance curve is almost never linear, in any market. Is the Alexa 8x the camera? Is a Mont Blanc ~100x better than a store brand pen? Is a Cartier worth so much more than a Timex? Is a 12-core Mac Pro 10x the computer of a basic Dell? The answer to every question depends on the individual and what they are looking for out of the product.

So it's difficult to say whether one camera is worth so much more than another, especially when the measure of the camera's worth is so nebulous and so subjective. What would make it worth 4x the cost? 8K resolution? 18 XLR inputs? A MF sensor? ProRes output? Even then, plenty of very smart and professional people have made a considered choice to purchase a Scarlet as it is - so it must be worth it to them. One great thing about the times is that there are plenty of options for everyone to choose from.

The difference from SD to HD is 4x and it is striking to anyone with functioning eyeballs. In this case my guess is that you would have to look very close.

Well, the difference from 480p to 1080p is actually 6.75x concerning the spatial resolution, whereas the difference from 1080p to 4096x2160 is 4.26x. So if you are looking for the same mathematical "wow factor" in the jump from ~2K to 4K as from SD to ~2K, it's just not in the numbers, even given the best case scenario. That being said, there has been a lot of praise for 4K projection, even from people who were naysayers prior to their first 4K projection experience.

In the end, the "4k" Red One looked softer than the 1.9k F35. Why? I can't say for sure, but in the end we do not shoot charts and our eyes tell us if it is good or not.

Interesting...many people at that time were saying that the RED One was too sharp (which really meant it had too much resolution, since it doesn't sharpen in camera), and that they preferred the F35s lower resolution coupled with electronic sharpening, which they felt produced a more film-like image.

Comparing output resolution of 4k to 1.9k by the numbers makes no sense because any image can be scaled to fit.

I agree with this statement very strongly. I really think cameras should be labelled not by the number of pixels on their chip, or the number of pixels in the file they output, but by the actual maximum resolved detail they get from a very repeatable and accurate test. Point everything at a standard zone plate, or better yet RGB zone plates. That'll get some hard fact into those numbers that can't really be avoided. And sure, things can be sharpened or softened in production and/or in post; focus, filters, and lighting are effective, grain can be added or removed, contrast can be adjusted. So is it possible to make a film on an FS-100 that appears sharper than a Scarlet, Epic, F3, Alexa, etc? Sure - but it wouldn't really be a fair assessment as those others would have to have been crippled in some way.

 

[rx8pilot]

Define "real" resolution. The number of pixels on the sensor? In that case, the Scarlet is definitely a 4K camera, and the FS-100 is probably in excess of a 2.88K camera. The number of pixels in the output image? In that case, the Scarlet is definitely a 4K camera, and the FS-100 is a 1080p camera. Or the number of pixels resolved under a standardized set of circumstances? Any mosaic pattern single-sensor camera won't resolve the full resolution of it's sensor. So in that case, the Scarlet is (conservatively) a 2.5-3.2K camera and the FS-100 a 1.6K camera. Any way you want it, that's pretty apples to apples.

I had a loooong phone call with Graeme Nattress at RED about this yesterday. The definition of resolution is not terribly difficult to express, but the resolution itself in terms of resolving power of an imaging system is difficult to express. When looking at a zone plate, the image must travel through the entire path before it can be measured. This takes all variables into account including the chart itself, its lighting, the lens on the camera and operators ability to nail the focus. The list goes on and on, neither one of use came up with any useful solution to comparing the resolution of multiple cameras other than our eyes. The zone plate is a tough test for sure, but if the manufacturer chooses, the system can be optimized for that test which can be mis-leading as well. There is no industry agreement about how Bayer resolution translates to MTF - it is too complicated. We must rely on our eyes in the end, which is what I ultimately do (along with many others). Resolution IMO, is about 25% of the image. Look at Alexa, it is AWESOME and very successful. It is also the lowest res camera in its class and I would shoot with it over a Scarlet any day of the week.

To be fair, however, the price/performance curve is almost never linear, in any market. Is the Alexa 8x the camera? Is a Mont Blanc ~100x better than a store brand pen? Is a Cartier worth so much more than a Timex? Is a 12-core Mac Pro 10x the computer of a basic Dell? The answer to every question depends on the individual and what they are looking for out of the product.

Agreed, I have written the checks to prove that. My point was more directed at those looking at their FS100 wondering if they should have purchased a Scarlet. I look at it from a pure business perspective, can your clients and audience tell the difference? Are they willing to pay more for that difference? Do they want a trendy camera on the set because it makes them feel better about themselves? At the end of the day, is the final product going to benefit from spending 4x on the camera that moves you up on a barely perceptible amount in terms of overall quality? Most of the jobs I have worked on have been no expenses spared type if things, so I am spoiled rotten. As I sit there shooting with an Alexa or F65, I can't help but notice how good the FS100 is at $5k. The camera is worth what people are willing to pay. The F35 original price was over $300k with SR-PC1 and other options - not many takers on that one.

Interesting...many people at that time were saying that the RED One was too sharp (which really meant it had too much resolution, since it doesn't sharpen in camera), and that they preferred the F35s lower resolution coupled with electronic sharpening, which they felt produced a more film-like image.

The F35 has 5760 (from memory) horizontal pixels and needs no sharpening to look its best. In general, MANY people say HD images (even bad ones) are "too sharp". Obviously that is subjective and does not make a good argument. For the purpose of this test, I was only concerned with making my particular F35 look it's best in the given circumstances. When A/B compared the RED looked softer, not just to the F35, but also the Viper, HPX-3000, and the F23. The D21 was about the same but with much better dynamic range and colorimitry at the time. 35mm film was our baseline. A whole pile of ASC members and industry leaders in both technical and creative positions made similar observations. RED has improved considerably since then with software upgrades which goes to show that the sensor is only the beginning - EVERYTHING must work together.

I agree with this statement very strongly. I really think cameras should be labelled not by the number of pixels on their chip, or the number of pixels in the file they output, but by the actual maximum resolved detail they get from a very repeatable and accurate test. Point everything at a standard zone plate, or better yet RGB zone plates. That'll get some hard fact into those numbers that can't really be avoided. And sure, things can be sharpened or softened in production and/or in post; focus, filters, and lighting are effective, grain can be added or removed, contrast can be adjusted. So is it possible to make a film on an FS-100 that appears sharper than a Scarlet, Epic, F3, Alexa, etc? Sure - but it wouldn't really be a fair assessment as those others would have to have been crippled in some way.

I have ben pushing for standardized tests for a decade and had nothing but frustration. There is no cooperation from manufacturers, or the industry in general. The test themselves are very difficult to conduct and far more difficult to conduct in a distributed environment. There are too many ways to mess up the tests. The only thing I have come up with is an independent group gathering all current cameras and doing it, but who pays? The ASC test I did involved 250 people, 2 days of shooting/testing, a few weeks of pre-pro, months of post and presentation prep. It was crazy - on top of that we were criticized about how the tests were done - that they favored certain cameras. Crazy.

 

[Rick Burnett]

I'd say a big problem with the tests is that many different things are trying to be measured at the same time, and how they play into each other is complicated. You've got both horizontal AND vertical resolution. You've got sensors rotated 45 degrees basically in terms of pixels. You've got color resolution differences that can be very different. I spent a few days talking with Barry Green over some charts he shot on the FS100 and the AF100. To me, at least, it clearly showed more color resolution on the AF100. Yet, owning both of them at one point, I never really saw the difference in my footage. Even knowing what the charts had said.

Same thing with Moire in the FS100. I've yet to get it in a shot, and believe me, I've looked. Yet, on a chart, I see it no problem. Don't misunderstand, I think charts are important, what I have trouble with is the translation into real world examples. At what point of moire in the resolution chart does it become a concern. I think that is what a lot of people scratch their heads with.

If you look at the FS100 and the Scarlet-X (extrapolating from the Epic examples since no one has done this yet) the Scarlet-X without question is going to shoot a MUCH nicer chart. But, given a chart is done correctly, it represents the absolute WORST scenario for ANY camera in terms of spatial frequency, but, it also represents the entire range of spatial frequency.

What I want to know, have people don't studies as to how spatial frequency exists in shooting? And what percentage of it matters? Probably not because that is HUGELY complicated, and very subjective, but, I think that's really the problem and why some people get frustrated at pixel peepers who care about those extreme conditions, and why pixel peepers get frustrated at those that don't look, because they probably want the best image ever and to not have to think about it.

Put yourself in the shoes of a DSLR shooter who has seen the charts for say the 7D (which I own as well). You see all this horrible moire. Now, you look at the FS100 charts and you see moire as well. Many people, because I've heard them in conversations, assume that the FS100 suffers from moire just like the Canons. Anyone who has shot on both KNOWS this isn't the case. I've shot quite a bit of stuff and never seen it pop out, even though I do know at extreme spatial frequency, I could see it. But I haven't yet, and again, I've tried.

THIS is the problem. These tests don't have an easy way of translating into what exactly is a usable range of usage where those issues fall off on the chart.

I evaluated the Scarlet-X in terms of moving up from the FS100. For me, it's just not worth the investment right now. I love the low light of the FS100, the battery life which is amazing, the weight, the low cost media. I'd love the improved rolling shutter if it is indeed as fast as Jim says it is, but when I put all that down on the table, the resolution just isn't as important for me. I shoot mainly content that gets released on our YouTube channel and we make and sell DVDs with the content. That is not to say I don't think the Red is a fantastic camera, I have no doubt it's IQ surpasses my FS100 in almost every way, but when you look at the WHOLE equation of our production, it doesn't make sense.

SO many variables.

 

[rx8pilot]

Rick,

Well said. We do not shoot charts for a living, we tell stories. We communicate emotions with moving images. if a camera test is done well, it will provide the intended audience some actionable feedback on what to expect when they are shooting real things. As you pointed out, the various charts stress the camera systems beyond what we experience in most of the real world. The only thing that can make a usable correlation between charts and real images is a good set of eyeballs and a human brain. I keep a huge set of charts that I use on all cameras that I come across. I have become familiar with them in terms of what I see in the chart and how that will likely manifest itself into the final pictures. The ugliest stuff in the charts does not always float to the surface but it is relative. The 7D is quite awful on a zone plate, but we have all seen some good stuff from these cameras. The DP's shooting with them have learned how to avoid its limitations not the least of which is shallow DOF. A creamy out of focus background wont show aliasing issues because everything is out of focus (but it will show banding, another limitation). The FS100 shows less moire on a chart, but it is still there of course only at higher frequency. In the real world it is more pleasing in the details (IMHO of course), go figure. The chart is helpful in getting a relative prediction of which camera will fall apart first in a given circumstance. This of course is true for many tests - not getting a numerical result but rather a comparative result. Resolution and dynamic range are fairly easy to compare, but not colorimetry which is the most subjective of them all and nearly impossible to chart or numerically quantify. For each manufacture it is either the secret sauce or the nail in the coffin. The Alexa shows this very well. It is a GORGEOUS camera, but there is no chart that would tell me that - put some people in the frame with some nicely shaped lighting and it looks great with very little effort - but it costs $65k+ to get that.

Yes, there are MANY variables. It seems that the biggest one is always cost. "I want the BEST possible camera! (that I can afford)" Science and subjective tests don't matter if you cannot afford it or it otherwise makes no financial benefit.