Web Analytics
Re: Sony FX9 camera test

Re: Sony FX9 camera test


alister@...
 

Hi Alfonso.

This agrees with my own findings. A touch more DR at 320 than 800. For me though the biggest difference becomes visible during practical shooting where at 320 ISO the final highlight roll off is visually slightly more pleasing than at 800 where the cut-off is sharper. The range isn’t greatly different when measured, but the smoothness is better at 320.

But as you point out these are small differences and not really something that will mean one shot is useable and the other not, just from a personal perspective I would try to use 320 ISO rather than 800 where possible as the lower noise and smoother highlights are just that little bit better. It gets even more confusing if you try to find the optimum settings on the FX3 as there is very little guidance on which ISO is 0dB and you can very easily end up at -12dB in some circumstances (with strong black clipping). And if you do ever play with the Alphas and FX3 be aware the ISO ratings are 1.3 stops different to the pro cameras for the standard gamma curves, but the same for the log gamma curves, I have no idea how or why this should be.

Camera tests are great learning exercises but even the most meticulous tests can miss things that don’t show up until the camera is put to use. For example I spent a lot of time comparing the FX6, FX9 and Venice and broadly everything was as expected, Venice is that little bit better. As you discovered more even noise in each channel,  better colour consistency, fewer processing artefacts and a touch more useable DR. I found the FX6 and FX9 to be quite similar in some areas but different in others. They have very different fixed pattern noise with the FX6 showing horizontal noise bands and the FX9 showing a fair amount of smear in high contrast but low light. But one thing I completely missed, even though I looked at various zone plates was that the FX6 is much more prone to coloured moire than the FX9 or Venice with very fine red and blue textures. Next time I need to add R, G and B zone plates to my tests. This difference wasn’t found until real world filming and I just happened to be shooting fabric with the right texture at the right distance. Now I’ve seen this I can reproduce it easily, but I didn’t spot it during my testing. It does also explain one discovery in the test which was that even though the FX9 has a 6K sensor it did not resolve significantly more than the 4.2K FX6 sensor and the Imatest plots didn’t look greatly different. But now I know the FX6’s OLPF is not as well optimised as the FX9’s.

This is a wonderful time to be a cinematographer. There are so many choices. There are lower cost but great performing cameras for personal projects or low budgets and the more expensive flagship cameras that are just that little bit better when the budget will allow. 

But whether you are using the budget camera or the flagship camera it really is going to be the skill of the operator that will make the greatest difference. Material shot well with an FS7 will look better than badly shot Venice material (of course this is not a surprise). From the audiences perspective the well shot FS7 (or other lower cost camera) should not look deficient, they will not know or care that it was shot on an FS7. As an owner of an FX9 I feel it is a great camera. But if a job came in tomorrow where the budget or some other aspect meant that Venice or an LF was more appropriate, then I would have no hesitation moving to the better camera, why wouldn’t you? But at the same time I know that if I had to cut in some FX3 crash cam or some other shot that could only be done with a small, light camera, from the audiences perspective they should not be aware that the “lesser” camera was used. Of course there are differences from budget to high end, but not having a high end camera is no longer an excuse for not producing incredible looking content.


Alister Chapman 

Cinematographer - DIT - Consultant
UK Mobile/Whatsapp +44 7711 152226


Facebook: Alister Chapman
Twitter: @stormguy



www.xdcam-user.com    1.5 million hits, 100,000 visits from over 45,000 unique visitors every month!  Film and Video production techniques, reviews and news.


















On 23 Apr 2021, at 23:19, alfonso parra <info@...> wrote:

Alister, I have tested the FX9 camera with the S-cinetone curve with the value 320 ISO / 0Db

In the analysis of the Stouffer scale, the total RD with a value of 320 ISO is 9.37 stop and with a ISO 800 it is 9.78, that is, approximately a little more than 1/3 of a stop with this last value of ISO. Now, if we consider a certain amount of noise, for example, the medium value, which is the one that usually coincides with other RD tests, at 320 ISO is 9.97 stops, while with a value of 800 ISO it is 9.21, around 2/3 more stop. The average SNR for example in Y is 48.6, while at a value of 800 ISO it is 46.2, that is, an increase of 2.4 Db. The best signal-to-noise ratio is clear at 320 ISO which allows for that greater dynamic range. Now, I wanted to contrast that difference with a black and white texture chart to assess where I am missing the detail. In whites at 320, a little more detail is observed in the highlights, up to 4 1/3, while with 800 at that value there is already a slight loss of detail in some areas of the textures, although it is not yet clipping. in such a way that I would consider that the real difference between using those two ISO in the highlights is slightly more than 1/3 of a stop at 320 ISO. In the shadows the range really remains more or less the same, with 5 stops of detail, although at those 5 stops below the middle gray I have observed a little more noise at ISO 320. Actually, it seems to me that the range Dynamic at 800 "slides" down relative to 320 ISO and that shift could be over 1/3 stop. The difference with respect to the S-log3 curve seems to me to be still similar. Evaluating the medium gray in relation to the white of 90%, the medium gray is located as you indicated at 44% A value that is within the values ​​relative to the STD curves.

If I consider the RD of the S-cinetone at 320 ISO it would be 4 above the medium gray and between 4 ½ and 5 below. If I consider the RD of the S-cinetone at 800 ISO it would be between 3 ½ or 3 2/3 and five below. The differences are not visually noticeable and it is necessary to go into detail to be able to see them. If an increase in sensitivity with the S-Cinetone curve is needed due to the existing light conditions, it is possible to work at 800 considering that small loss of RD in the highlights. The noise level in the shadows is good enough not to have an observable loss of image quality.

 I really don't know if the differences we can find are worth it, but I think that taking the tests is a learning process, at least that's how I understand it, and it serves to understand, compare and ultimately to decide. It does not matter so much if the differences are great or not, but the knowledge itself that is obtained.

At present I believe that there are no great differences, nor great discoveries when we do camera tests and you have to go into a lot of detail to observe them. A famous Colombian boxer used to say that “it is better to be rich than poor”, and paraphrasing it we can say that it is better to shoot with a good camera than with a bad one, or that it is preferable to shoot with a very good camera rather than with a good camera, but The truth is that the difference in the image between a good camera and a very good camera is invisible to the viewer, and the difference will be in the look of the cinematographer, although it is true that improvements in cameras sometimes have no impact both in the image quality but in the options that the cinematographer has to work with. For example, with the FS7 excellent images are obtained as with the FX9, but with this we have the Dual ISO which is very practical.

In the end, we learn and learn as Art says in a bottomless pit… fortunately.

 Let me know Alister if these results agree with your tests

Regards

Alfonso Parra ADFC
www.alfonsoparra.com
Tel Colombia 57 311 5798776
Tel Spain 34 639109309
Instagram alfonso_parra_adfc




Join cml-raw-log-hdr@cml.news to automatically receive all group messages.