Tribute to a7s III

[jonpais]

https://youtu.be/SJZvEe8oM6k

For a long while now, we've been keenly aware that highlights in skin tones and light-colored fabrics have been difficult if not impossible to recover in post, leading us to re-examine our ETTR method. After watching a video by Gerald Undone, we tried using two zebra settings, aiming for an ETTR of only +1.66 stops - well below what we were accustomed to. However, whereas using zebras had for years been liberating, freeing us from the drudgery of having to schlep around a monitor just for the waveforms, the new routine was beginning to make shooting an unbearable drudgery, prompting us to write only a couple weeks ago

Using zebras can be a real nuisance! Let’s say that in the menus, I’ve got zebras Custom 1 set to 55% for 18% gray (for an ETTR of +1.66 stops) and Custom 2 Lower Limit set to 94% (the limit before clipping occurs with S-Log3). Each time the lighting conditions change, I’ve got to pass the X-Rite ColorChecker to the model, take a reading of the gray card, then dive into the menus and switch to Custom 1 to check for highlight clipping. However, because it’s an exercise in futility trying to see zebras with Sony’s ludicrously subpar low resolution LCD, it’s necessary to engage magnification; but wouldn’t you know, it’s no longer possible to adjust zebras while the image is magnified, necessitating doing this dance between the magnifier and zebras – by which time the light has probably changed and the model has collapsed out of despair.

Anyhow, words can't express the exhilaration we felt - after a half year of setbacks - on discovering the secret to taming those highlights and extracting the full 12-1/2 tasty stops of dynamic range out of our a7s III.

 

[DLD]

And now there are rumors of A7SIV out. 40 MPX 8K. No prices are given but I assume A7SIII will slide down to ~ $3,000 with SIV getting into the $4,500 range. Canon R5 is doing extremely well in terms of sales. Sony must want some of that market and will respond with similar specs at a similar price but without the similar overheating.

 

[ahalpert]

And now there are rumors of A7SIV out. 40 MPX 8K. No prices are given but I assume A7SIII will slide down to ~ $3,000 with SIV getting into the $4,500 range. Canon R5 is doing extremely well in terms of sales. Sony must want some of that market and will respond with similar specs at a similar price but without the similar overheating.

That rumor is complete BS. As I recall, it was posted with little confidence and just for fun. But consider how long it took for them to release the A7SIII after the a7sii. And also, the rumor said that this high res A7SIV would have the same high ISO performance as the A7SIII. Which is not possible

 

[DLD]

That rumor is complete BS. As I recall, it was posted with little confidence and just for fun. But consider how long it took for them to release the A7SIII after the a7sii. And also, the rumor said that this high res A7SIV would have the same high ISO performance as the A7SIII. Which is not possible

Eh, depending on noise suppression vs. the resolution loss on a higher MPX sensor, it's possible with a compressed codec. Which it will be in 8K.

As to the commercial reasons, A7SIII could have come out several years earlier, albeit in a different form (and at a different price). It's not that Sony lacked the engineering expertise. It was a corporate decision, for the lack of a better term. R5, however crippled, busted the previously solemn obedience to the establishment. A1 came out within half a year.

But having completed the top spec challenge, it also makes sense for Sony to match R5 on video, without the costs associated with the high frame rate photography in a similar price bracket. A1 was a statement* but it's almost three grand above R5. Beside that, I think A7SIII is somewhat overpriced and could easily use a price drop similar to A7RIV (from $3,500 to $3,000) within a year and a half from its initial release date.

*In that Russian language interview, the Canon poobah was asked about R3 being a response to Sony and his reply was something along the lines of "We don't know what Sony's plans are". But I think they do via their industry sources, at least, half a year in advance. Everyone does. So, if R3 - sports/action/BiF/high res camera - is a response to A1, A7SIV will be a real response to R5. The confusion, if any, is that it will be the 7S-line lower res/high ISO model. But the model classification is likely irrelevant in the grand scheme of things. It's just that Sony's stuck with A9, A7S, A7R, A7 model naming structure.

 

[DLD]

Also, with Canon planning to offer a lineup of "affordable" 8K cameras, Sony will have to respond there as well. And that probably means not only the 8K A7SIV but 8K FX-IV, 8K FX-6 II and 8K FX-9 II as well. FX-9 II might have a different 8K sensor though. Or not.

 

[Trek of Joy]

https://youtu.be/ZFODtS431Yg

For a long while now, we've been keenly aware that highlights in skin tones and light-colored fabrics have been difficult if not impossible to recover in post, leading us to re-examine our ETTR method. After watching a video by Gerald Undone, we tried using two zebra settings, aiming for an ETTR of only +1.66 stops - well below what we were accustomed to. However, whereas using zebras had for years been liberating, freeing us from the drudgery of having to schlep around a monitor just for the waveforms, the new routine was beginning to make shooting an unbearable drudgery, prompting us to write only a couple weeks ago

Using zebras can be a real nuisance! Let’s say that in the menus, I’ve got zebras Custom 1 set to 55% for 18% gray (for an ETTR of +1.66 stops) and Custom 2 Lower Limit set to 94% (the limit before clipping occurs with S-Log3). Each time the lighting conditions change, I’ve got to pass the X-Rite ColorChecker to the model, take a reading of the gray card, then dive into the menus and switch to Custom 1 to check for highlight clipping. However, because it’s an exercise in futility trying to see zebras with Sony’s ludicrously subpar low resolution LCD, it’s necessary to engage magnification; but wouldn’t you know, it’s no longer possible to adjust zebras while the image is magnified, necessitating doing this dance between the magnifier and zebras – by which time the light has probably changed and the model has collapsed out of despair.

Anyhow, words can't express the exhilaration we felt - after a half year of setbacks - on discovering the secret to taming those highlights and extracting the full 12-1/2 tasty stops of dynamic range out of our a7s III.

Great colors. What's the secret to taming the highlights you mentioned? I've been using the zebras, but I'm always open to trying something else that's working. I shoot in bright sun a lot and keeping stuff from clipping can be a PITA at times.

Cheers

Chris

 

[jonpais]

Thanks, Chris. For all I know, this might only apply to HDR (that’s all I ever shoot!), but the secret is to (a) shoot ProRes RAW (HQ); (b) use the false color of the Ninja V as a guide; (c) avoid the false colors bright orange and red; (d) transcode to ProRes 4444 in Apple Compressor; then (e) grade in DaVinci Resolve Studio. The shadow, midtone and highlight wheels of Final Cut Pro are unsuitable for HDR grading. Do not shoot S-Log3 4:2:2.

If on the other hand, you aren’t shooting HDR; you don’t use a Ninja V; and you just shoot XAVC internally - try using the magnifier. I found that zebras often escape detection with the low resolution LCD of the a7s III and I can see them again with the aid of the magnifier. The downside being, that I can’t adjust the zebras while the magnifier is engaged.

If you ETTR like I do, you might also find that +3 stops isn’t always necessary. Gerald Undone prefers +1.66 stops, setting C1 to 55 for 18% gray then checking the highlights with C2 Lower Limit at 94%. But I found that whole process too insane for my style of shooting.

Finally, I back off on exposure more when shooting people, so using false color, a little yellow is okay in the highlights, but I wouldn’t allow the entire face to be covered in yellow.

Hope that helps!

Edit: I should have said that Gerald’s settings are for S-Log3.