p.4 #1Very frustrated by Sigma 45mm 2.8 autofocus problems
AGeoJO wrote: chiron wroteAGeoJO wrotechiron wroteAGeoJO wrote@chiron@@@@@@ - Peter, I just checked that lens on my A1, just a few shots and they seem to focus fine in AF-C from within a 2-foot distance or so indoors. I will try some again tomorrow as the light is winding lower now.That is good news, Joshua. Thank you for checking that for me. The A1 is a lot of firepower to throw at a lens to make it focus.Peter,
I took 15 images using the AF-C and Eye-AF of my granddaughter as she walked towards me on the A1. The drive was set at 5 frames per second. She started out some 8 feet away and ended up some 2 feet from me. Seven out of the 15 are in focus and 8 are not in focus. The sequence from far to close is as follows: 3 in focus, 3 not in focus, 2 in focus, 3 not in focus, 1 in focus, 2 not in focus and 1 in focus.
For single shots of people moving around in AF-C, all are in focus at any distance and I took images in a single drive using Eye-AF.
I wouldn't use the lens for fast moving/walking target, especially in a burst but it seems to be fine for single exposure drives even in AF-C. Again, using the A1 and after the most recent FW upgrade. Your mileage may vary, however.Joshua,
That's very informative. Thank very much you for doing it.
The lens on an A9 does seem to be fine on stationary objects in AF-C, but for me the tracking focus with eye-af seems so unreliable as to not make the rendering loveliness of the lens worth the high percentage of ruined images.
One clarification--what were the settings and shooting situation with the people walking around? Were you not using eye-af on them? Was eye-af turned on? Was it all single shots of people who were still or were they actually in motion when you made the single shot? Tracking seems to be one of the questionable operations with this lens.
It does sound possible that the A1 vs the A9 is boosting the performance of this difficult lens.Peter, I used it indoors but there was plenty of light. I set the shutter speed at 1/60 sec at f/2.8 and auto ISO. Yes, the Eye-AF was on and I could see that on the viewfinder, even on the out-of-focus images when I was using the continuous drive. On the single shot mode, my grandkids didn’t pose while playing and they still moved around but their movement is definitely slower than typical walking speed in terms of distance changes and the lens didn’t have any problems there. That lens had only an approximately 50% success rate of tracking slowly walking kid and that using the A1. I still like the rendition of that lens though and I would not hesitate of using that lens for street photography. And, I know its limitation now....Show more →
Thank you, Joshua.
If you were using it for street photography or travel, which are two of the situations in which I would like to use it, would you set it for AF-S and work within that limitation, or would you try to use it in AF-C because you would be at a much great distance? Or use it in AF-C on stationary or slow moving targets?
p.4 #2Very frustrated by Sigma 45mm 2.8 autofocus problems
BlueBomberTurbo wrote: Teo Rey wrotechiron wrotefreaklikeme wrotechiron wroteExperimenting with the autofocus a bit. Some of the things I have tried and noticed.
1. Real Time Tracking in af-c seems to be working well enough. I have tried locking focus and tracking objects in dim light at about 10 feet and at 1.5 feet (at the short distance there was a high contrast target) by moving the camera erratically and quickly. In both cases the, the lens on the A9 maintained focus despite the dim light and the quick erratic movements. When I took a photo at the end of the erratic tracking, the image was sharp within the limits of this slightly soft lens. All this was done at f2.8.
2. I put up an image of Mona Lisa on the screen of my lap top and used eye-af from about 20 or so inches. The eye-af found her eye quickly and stayed locked when I moved the camera erratically. I then dimmed the brightness of the screen quite a bit and the camera still stayed locked. In both cases, the resulting image was sharp (within the sharpness limitations of the lens).
It is probably true that the Mona Lisa on the computer screen, even when dimmed quite a bit, provides more contrast and brightness, because its light is projected, than a real object would provide since the real object is seen in less intense reflected light. It is also a two-dimensional tracking target, which I suspect would be easier for the processor to compute than would be a three-dimensional object. But it was interesting to me that the lens and A9 did well with tracking the the Mona Lisa's eye, even when jittering around and at a close distance.
Some of this makes me think there is a chance the lens will perform better on an A1 because of the increased calculations. Anyone have this combination to try out?I think you're probably seeing good results with Mona because she's two dimensional and not moving.Yes, although I did jitter the lens around while focus was locked on her eyes and it seemed to hold and produce a focused image. Still, the screen is two-dimensional and probalably has greater contrast than most reflected light scenes.
But there may be something to the idea that the lens isn't processing movement fast enough or well enough (which I guess doesn't say much since that is the whole point of af-c). itai195 upthread points out that in a series of shots the first is focused and the others are not and if you start a new series the same thing happens. So af-c acquires focus, presumably using pdaf, and then can't track movements while maintaining focus. But since it does acquire focus in the first place it seems like this should be a fixable problem unless their focus motor or data throughput circuitry can't handle the speed required. But it seems more likely to be a bad algorithm.I think it sounds like @BlueBomberTurbo@@ found a great solution for a poor focusing Samyang lens, the 50mm f1.4, in another thread. Perhaps his solution would work here too?Haha, guess we were typing at the same time!
And yes, the difference is amazing. Going from AF performance much worse than the worst adapted lens (probably my Yongnuo 35/2) to that of a very decent native lens, within the limitations of the lens' slower AF motor. Sigma should certainly have E-mount figured out by now, so maybe it's another AF motor incompatibility issue like on that first Samyang/Rokinon AF motor in the 50/1.4.
I'm going to try this with the Sigma 45 and I'll report back. Thank you!
What was the AF motor incompatibility issue with the first Samyang/Rok 50/1.4? And how did it manifest in problems? It is at least plausible that something similar is going on with Sigma since their 45 acquires focus in AF-C but then has trouble following for tracking. Also, they have AF problems scattered through their e mount lenses.
p.4 #3Very frustrated by Sigma 45mm 2.8 autofocus problems
chiron wrote: AGeoJO wrotechiron wroteAGeoJO wrotechiron wroteAGeoJO wrote@chiron@@@@@@@ - Peter, I just checked that lens on my A1, just a few shots and they seem to focus fine in AF-C from within a 2-foot distance or so indoors. I will try some again tomorrow as the light is winding lower now.That is good news, Joshua. Thank you for checking that for me. The A1 is a lot of firepower to throw at a lens to make it focus.Peter,
I took 15 images using the AF-C and Eye-AF of my granddaughter as she walked towards me on the A1. The drive was set at 5 frames per second. She started out some 8 feet away and ended up some 2 feet from me. Seven out of the 15 are in focus and 8 are not in focus. The sequence from far to close is as follows: 3 in focus, 3 not in focus, 2 in focus, 3 not in focus, 1 in focus, 2 not in focus and 1 in focus.
For single shots of people moving around in AF-C, all are in focus at any distance and I took images in a single drive using Eye-AF.
I wouldn't use the lens for fast moving/walking target, especially in a burst but it seems to be fine for single exposure drives even in AF-C. Again, using the A1 and after the most recent FW upgrade. Your mileage may vary, however.Joshua,
That's very informative. Thank very much you for doing it.
The lens on an A9 does seem to be fine on stationary objects in AF-C, but for me the tracking focus with eye-af seems so unreliable as to not make the rendering loveliness of the lens worth the high percentage of ruined images.
One clarification--what were the settings and shooting situation with the people walking around? Were you not using eye-af on them? Was eye-af turned on? Was it all single shots of people who were still or were they actually in motion when you made the single shot? Tracking seems to be one of the questionable operations with this lens.
It does sound possible that the A1 vs the A9 is boosting the performance of this difficult lens.Peter, I used it indoors but there was plenty of light. I set the shutter speed at 1/60 sec at f/2.8 and auto ISO. Yes, the Eye-AF was on and I could see that on the viewfinder, even on the out-of-focus images when I was using the continuous drive. On the single shot mode, my grandkids didn’t pose while playing and they still moved around but their movement is definitely slower than typical walking speed in terms of distance changes and the lens didn’t have any problems there. That lens had only an approximately 50% success rate of tracking slowly walking kid and that using the A1. I still like the rendition of that lens though and I would not hesitate of using that lens for street photography. And, I know its limitation now.Thank you, Joshua.
If you were using it for street photography or travel, which are two of the situations in which I would like to use it, would you set it for AF-S and work within that limitation, or would you try to use it in AF-C because you would be at a much great distance? Or use it in AF-C on stationary or slow moving targets?
Unless the target is 100% stationary or completely non-moving, my preference is to set my AF mode to AF-C. Even then, I prefer AF-C although the drive can be a single shot. Since I am using the back button AF, even for landscape, I use the AF-C pretty much all the time.
p.4 #4Very frustrated by Sigma 45mm 2.8 autofocus problems
chiron wrote: BlueBomberTurbo wroteTeo Rey wrotechiron wrotefreaklikeme wrotechiron wroteExperimenting with the autofocus a bit. Some of the things I have tried and noticed.
1. Real Time Tracking in af-c seems to be working well enough. I have tried locking focus and tracking objects in dim light at about 10 feet and at 1.5 feet (at the short distance there was a high contrast target) by moving the camera erratically and quickly. In both cases the, the lens on the A9 maintained focus despite the dim light and the quick erratic movements. When I took a photo at the end of the erratic tracking, the image was sharp within the limits of this slightly soft lens. All this was done at f2.8.
2. I put up an image of Mona Lisa on the screen of my lap top and used eye-af from about 20 or so inches. The eye-af found her eye quickly and stayed locked when I moved the camera erratically. I then dimmed the brightness of the screen quite a bit and the camera still stayed locked. In both cases, the resulting image was sharp (within the sharpness limitations of the lens).
It is probably true that the Mona Lisa on the computer screen, even when dimmed quite a bit, provides more contrast and brightness, because its light is projected, than a real object would provide since the real object is seen in less intense reflected light. It is also a two-dimensional tracking target, which I suspect would be easier for the processor to compute than would be a three-dimensional object. But it was interesting to me that the lens and A9 did well with tracking the the Mona Lisa's eye, even when jittering around and at a close distance.
Some of this makes me think there is a chance the lens will perform better on an A1 because of the increased calculations. Anyone have this combination to try out?I think you're probably seeing good results with Mona because she's two dimensional and not moving.Yes, although I did jitter the lens around while focus was locked on her eyes and it seemed to hold and produce a focused image. Still, the screen is two-dimensional and probalably has greater contrast than most reflected light scenes.
But there may be something to the idea that the lens isn't processing movement fast enough or well enough (which I guess doesn't say much since that is the whole point of af-c). itai195 upthread points out that in a series of shots the first is focused and the others are not and if you start a new series the same thing happens. So af-c acquires focus, presumably using pdaf, and then can't track movements while maintaining focus. But since it does acquire focus in the first place it seems like this should be a fixable problem unless their focus motor or data throughput circuitry can't handle the speed required. But it seems more likely to be a bad algorithm.I think it sounds like @BlueBomberTurbo@@@ found a great solution for a poor focusing Samyang lens, the 50mm f1.4, in another thread. Perhaps his solution would work here too?Haha, guess we were typing at the same time!
And yes, the difference is amazing. Going from AF performance much worse than the worst adapted lens (probably my Yongnuo 35/2) to that of a very decent native lens, within the limitations of the lens' slower AF motor. Sigma should certainly have E-mount figured out by now, so maybe it's another AF motor incompatibility issue like on that first Samyang/Rokinon AF motor in the 50/1.4.I'm going to try this with the Sigma 45 and I'll report back. Thank you!
What was the AF motor incompatibility issue with the first Samyang/Rok 50/1.4? And how did it manifest in problems? It is at least plausible that something similar is going on with Sigma since their 45 acquires focus in AF-C but then has trouble following for tracking. Also, they have AF problems scattered through their e mount lenses.
Nothing written about it, but it was Samyang's first attempt at an AF motor. A dual driven linear setup (2 motors pushing 1 focusing element), at that. It's much different than the motors that followed it (I've owned 5 other Samyang AF lenses). Even the dual driven linear ones in the 35 and 85 that followed it. It seems to be lacking in precision a bit, stepping further more noticeably in manual focus than any other lens I've used.
That said, AF-C was horribly random. Using Flexible Spot, Wide, Center, etc. led to AF being far off, in front and behind. Anything at a distance of more than maybe 10ft almost always failed. But what I noticed is that AF-S hit perfectly, and CDAF assisted settings (AF-S, DMF, AF-C + Eye/Face/Object tracking) were waaaaay more reliable than AF-C and PDAF alone. So I got to messing with settings, and found that changing AF Tracking Sensitivity from 5 (Responsive) to 3 (Standard) improved things greatly, and 1 (Locked on) made it act like any other first party lens.
The other thing was that this improvement was immediately visible in the EVF while focusing. At 5, the lens constantly jittered the focus in front and behind the subject, even after claiming to lock on. Whether it was miscommunication with the camera, allowing it to shoot when not completely in focus (I have both AF-C and AF-S Priority set to AF), or just the lack of precision in the motor, not allowing it to stay close enough to in focus while locked on, who knows. But setting AF Tracking Sensitivity down to 1 completely stopped these jitters, and stopped all of the focus issues in AF-C. The lens is now almost as good as it is in AF-S (which is near flawless).
p.4 #6Very frustrated by Sigma 45mm 2.8 autofocus problems
AGeoJO wrote: chiron wroteAGeoJO wrotechiron wroteAGeoJO wrotechiron wroteThat is good news, Joshua. Thank you for checking that for me. The A1 is a lot of firepower to throw at a lens to make it focus.Peter,
I took 15 images using the AF-C and Eye-AF of my granddaughter as she walked towards me on the A1. The drive was set at 5 frames per second. She started out some 8 feet away and ended up some 2 feet from me. Seven out of the 15 are in focus and 8 are not in focus. The sequence from far to close is as follows: 3 in focus, 3 not in focus, 2 in focus, 3 not in focus, 1 in focus, 2 not in focus and 1 in focus.
For single shots of people moving around in AF-C, all are in focus at any distance and I took images in a single drive using Eye-AF.
I wouldn't use the lens for fast moving/walking target, especially in a burst but it seems to be fine for single exposure drives even in AF-C. Again, using the A1 and after the most recent FW upgrade. Your mileage may vary, however.Joshua,
That's very informative. Thank very much you for doing it.
The lens on an A9 does seem to be fine on stationary objects in AF-C, but for me the tracking focus with eye-af seems so unreliable as to not make the rendering loveliness of the lens worth the high percentage of ruined images.
One clarification--what were the settings and shooting situation with the people walking around? Were you not using eye-af on them? Was eye-af turned on? Was it all single shots of people who were still or were they actually in motion when you made the single shot? Tracking seems to be one of the questionable operations with this lens.
It does sound possible that the A1 vs the A9 is boosting the performance of this difficult lens.Peter, I used it indoors but there was plenty of light. I set the shutter speed at 1/60 sec at f/2.8 and auto ISO. Yes, the Eye-AF was on and I could see that on the viewfinder, even on the out-of-focus images when I was using the continuous drive. On the single shot mode, my grandkids didn’t pose while playing and they still moved around but their movement is definitely slower than typical walking speed in terms of distance changes and the lens didn’t have any problems there. That lens had only an approximately 50% success rate of tracking slowly walking kid and that using the A1. I still like the rendition of that lens though and I would not hesitate of using that lens for street photography. And, I know its limitation now.Thank you, Joshua.
If you were using it for street photography or travel, which are two of the situations in which I would like to use it, would you set it for AF-S and work within that limitation, or would you try to use it in AF-C because you would be at a much great distance? Or use it in AF-C on stationary or slow moving targets?Unless the target is 100% stationary or completely non-moving, my preference is to set my AF mode to AF-C. Even then, I prefer AF-C although the drive can be a single shot. Since I am using the back button AF, even for landscape, I use the AF-C pretty much all the time.
Yes, that makes sense and is what I would try to do with the Sigma 45 on the street or traveling.
p.4 #7Very frustrated by Sigma 45mm 2.8 autofocus problems
BlueBomberTurbo wrote: chiron wroteBlueBomberTurbo wroteTeo Rey wrotechiron wrotefreaklikeme wroteI think you're probably seeing good results with Mona because she's two dimensional and not moving.Yes, although I did jitter the lens around while focus was locked on her eyes and it seemed to hold and produce a focused image. Still, the screen is two-dimensional and probalably has greater contrast than most reflected light scenes.
But there may be something to the idea that the lens isn't processing movement fast enough or well enough (which I guess doesn't say much since that is the whole point of af-c). itai195 upthread points out that in a series of shots the first is focused and the others are not and if you start a new series the same thing happens. So af-c acquires focus, presumably using pdaf, and then can't track movements while maintaining focus. But since it does acquire focus in the first place it seems like this should be a fixable problem unless their focus motor or data throughput circuitry can't handle the speed required. But it seems more likely to be a bad algorithm.I think it sounds like @BlueBomberTurbo@@@@ found a great solution for a poor focusing Samyang lens, the 50mm f1.4, in another thread. Perhaps his solution would work here too?Haha, guess we were typing at the same time!
And yes, the difference is amazing. Going from AF performance much worse than the worst adapted lens (probably my Yongnuo 35/2) to that of a very decent native lens, within the limitations of the lens' slower AF motor. Sigma should certainly have E-mount figured out by now, so maybe it's another AF motor incompatibility issue like on that first Samyang/Rokinon AF motor in the 50/1.4.I'm going to try this with the Sigma 45 and I'll report back. Thank you!
What was the AF motor incompatibility issue with the first Samyang/Rok 50/1.4? And how did it manifest in problems? It is at least plausible that something similar is going on with Sigma since their 45 acquires focus in AF-C but then has trouble following for tracking. Also, they have AF problems scattered through their e mount lenses.Nothing written about it, but it was Samyang's first attempt at an AF motor. A dual driven linear setup (2 motors pushing 1 focusing element), at that. It's much different than the motors that followed it (I've owned 5 other Samyang AF lenses). Even the dual driven linear ones in the 35 and 85 that followed it. It seems to be lacking in precision a bit, stepping further more noticeably in manual focus than any other lens I've used.
That said, AF-C was horribly random. Using Flexible Spot, Wide, Center, etc. led to AF being far off, in front and behind. Anything at a distance of more than maybe 10ft almost always failed. But what I noticed is that AF-S hit perfectly, and CDAF assisted settings (AF-S, DMF, AF-C + Eye/Face/Object tracking) were waaaaay more reliable than AF-C and PDAF alone. So I got to messing with settings, and found that changing AF Tracking Sensitivity from 5 (Responsive) to 3 (Standard) improved things greatly, and 1 (Locked on) made it act like any other first party lens.
The other thing was that this improvement was immediately visible in the EVF while focusing. At 5, the lens constantly jittered the focus in front and behind the subject, even after claiming to lock on. Whether it was miscommunication with the camera, allowing it to shoot when not completely in focus (I have both AF-C and AF-S Priority set to AF), or just the lack of precision in the motor, not allowing it to stay close enough to in focus while locked on, who knows. But setting AF Tracking Sensitivity down to 1 completely stopped these jitters, and stopped all of the focus issues in AF-C. The lens is now almost as good as it is in AF-S (which is near flawless). ...Show more →
Very interesting, Blue Bomber Turbo (hereafter BBT). I have only had a few minutes with the 45 on the camera and sensitivity set to 1, but it clearly does make a difference. I don't have another human at home at the moment, so I have had to use my golden retriever and one of the cats.
The preliminary verdict is that there is a lot of improvement, but it is still not quite what it should be. AF-C with eye-af activated now gets the hairs around the eye in sharp focus but the focus is not on the eyeball itself. Many of the pictures (of animals) would be useable, but the eyeball is still not sharp. I got the same results whether I was inside 2 feet or at 10 feet--this is also a change from when the sensitivity is set at 3.
This is a very interesting result. It does seem to make it clear that the problem here is with their algorithms. The lens can acquire focus on the hairs but is not distinguishing where the eyeball is.
p.4 #8Very frustrated by Sigma 45mm 2.8 autofocus problems
Chiron - Did you implement the very recent firmware update before doing this test? I ask because the firmware update has markedly improved accuracy in single shot autofocus so maybe it changes the algorithm that could also affect ability to focus on eyes vs. hair around eyes in continuous autofocus mode?
chiron wrote: BlueBomberTurbo wrotechiron wroteBlueBomberTurbo wroteTeo Rey wrotechiron wroteYes, although I did jitter the lens around while focus was locked on her eyes and it seemed to hold and produce a focused image. Still, the screen is two-dimensional and probalably has greater contrast than most reflected light scenes.
But there may be something to the idea that the lens isn't processing movement fast enough or well enough (which I guess doesn't say much since that is the whole point of af-c). itai195 upthread points out that in a series of shots the first is focused and the others are not and if you start a new series the same thing happens. So af-c acquires focus, presumably using pdaf, and then can't track movements while maintaining focus. But since it does acquire focus in the first place it seems like this should be a fixable problem unless their focus motor or data throughput circuitry can't handle the speed required. But it seems more likely to be a bad algorithm.I think it sounds like @BlueBomberTurbo@@@@@ found a great solution for a poor focusing Samyang lens, the 50mm f1.4, in another thread. Perhaps his solution would work here too?Haha, guess we were typing at the same time!
And yes, the difference is amazing. Going from AF performance much worse than the worst adapted lens (probably my Yongnuo 35/2) to that of a very decent native lens, within the limitations of the lens' slower AF motor. Sigma should certainly have E-mount figured out by now, so maybe it's another AF motor incompatibility issue like on that first Samyang/Rokinon AF motor in the 50/1.4.I'm going to try this with the Sigma 45 and I'll report back. Thank you!
What was the AF motor incompatibility issue with the first Samyang/Rok 50/1.4? And how did it manifest in problems? It is at least plausible that something similar is going on with Sigma since their 45 acquires focus in AF-C but then has trouble following for tracking. Also, they have AF problems scattered through their e mount lenses.Nothing written about it, but it was Samyang's first attempt at an AF motor. A dual driven linear setup (2 motors pushing 1 focusing element), at that. It's much different than the motors that followed it (I've owned 5 other Samyang AF lenses). Even the dual driven linear ones in the 35 and 85 that followed it. It seems to be lacking in precision a bit, stepping further more noticeably in manual focus than any other lens I've used.
That said, AF-C was horribly random. Using Flexible Spot, Wide, Center, etc. led to AF being far off, in front and behind. Anything at a distance of more than maybe 10ft almost always failed. But what I noticed is that AF-S hit perfectly, and CDAF assisted settings (AF-S, DMF, AF-C + Eye/Face/Object tracking) were waaaaay more reliable than AF-C and PDAF alone. So I got to messing with settings, and found that changing AF Tracking Sensitivity from 5 (Responsive) to 3 (Standard) improved things greatly, and 1 (Locked on) made it act like any other first party lens.
The other thing was that this improvement was immediately visible in the EVF while focusing. At 5, the lens constantly jittered the focus in front and behind the subject, even after claiming to lock on. Whether it was miscommunication with the camera, allowing it to shoot when not completely in focus (I have both AF-C and AF-S Priority set to AF), or just the lack of precision in the motor, not allowing it to stay close enough to in focus while locked on, who knows. But setting AF Tracking Sensitivity down to 1 completely stopped these jitters, and stopped all of the focus issues in AF-C. The lens is now almost as good as it is in AF-S (which is near flawless).Very interesting, Blue Bomber Turbo (hereafter BBT). I have only had a few minutes with the 45 on the camera and sensitivity set to 1, but it clearly does make a difference. I don't have another human at home at the moment, so I have had to use my golden retriever and one of the cats.
The preliminary verdict is that there is a lot of improvement, but it is still not quite what it should be. AF-C with eye-af activated now gets the hairs around the eye in sharp focus but the focus is not on the eyeball itself. Many of the pictures (of animals) would be useable, but the eyeball is still not sharp. I got the same results whether I was inside 2 feet or at 10 feet--this is also a change from when the sensitivity is set at 3.
This is a very interesting result. It does seem to make it clear that the problem here is with their algorithms. The lens can acquire focus on the hairs but is not distinguishing where the eyeball is. ...Show more →
p.4 #9Very frustrated by Sigma 45mm 2.8 autofocus problems
httivals wrote:
Chiron - Did you implement the very recent firmware update before doing this test? I ask because the firmware update has markedly improved accuracy in single shot autofocus so maybe it changes the algorithm that could also affect ability to focus on eyes vs. hair around eyes in continuous autofocus mode?
chiron wroteBlueBomberTurbo wrotechiron wroteBlueBomberTurbo wroteTeo Rey wroteI think it sounds like @BlueBomberTurbo@@@@@@ found a great solution for a poor focusing Samyang lens, the 50mm f1.4, in another thread. Perhaps his solution would work here too?Haha, guess we were typing at the same time!
And yes, the difference is amazing. Going from AF performance much worse than the worst adapted lens (probably my Yongnuo 35/2) to that of a very decent native lens, within the limitations of the lens' slower AF motor. Sigma should certainly have E-mount figured out by now, so maybe it's another AF motor incompatibility issue like on that first Samyang/Rokinon AF motor in the 50/1.4.I'm going to try this with the Sigma 45 and I'll report back. Thank you!
What was the AF motor incompatibility issue with the first Samyang/Rok 50/1.4? And how did it manifest in problems? It is at least plausible that something similar is going on with Sigma since their 45 acquires focus in AF-C but then has trouble following for tracking. Also, they have AF problems scattered through their e mount lenses.Nothing written about it, but it was Samyang's first attempt at an AF motor. A dual driven linear setup (2 motors pushing 1 focusing element), at that. It's much different than the motors that followed it (I've owned 5 other Samyang AF lenses). Even the dual driven linear ones in the 35 and 85 that followed it. It seems to be lacking in precision a bit, stepping further more noticeably in manual focus than any other lens I've used.
That said, AF-C was horribly random. Using Flexible Spot, Wide, Center, etc. led to AF being far off, in front and behind. Anything at a distance of more than maybe 10ft almost always failed. But what I noticed is that AF-S hit perfectly, and CDAF assisted settings (AF-S, DMF, AF-C + Eye/Face/Object tracking) were waaaaay more reliable than AF-C and PDAF alone. So I got to messing with settings, and found that changing AF Tracking Sensitivity from 5 (Responsive) to 3 (Standard) improved things greatly, and 1 (Locked on) made it act like any other first party lens.
The other thing was that this improvement was immediately visible in the EVF while focusing. At 5, the lens constantly jittered the focus in front and behind the subject, even after claiming to lock on. Whether it was miscommunication with the camera, allowing it to shoot when not completely in focus (I have both AF-C and AF-S Priority set to AF), or just the lack of precision in the motor, not allowing it to stay close enough to in focus while locked on, who knows. But setting AF Tracking Sensitivity down to 1 completely stopped these jitters, and stopped all of the focus issues in AF-C. The lens is now almost as good as it is in AF-S (which is near flawless).Very interesting, Blue Bomber Turbo (hereafter BBT). I have only had a few minutes with the 45 on the camera and sensitivity set to 1, but it clearly does make a difference. I don't have another human at home at the moment, so I have had to use my golden retriever and one of the cats.
The preliminary verdict is that there is a lot of improvement, but it is still not quite what it should be. AF-C with eye-af activated now gets the hairs around the eye in sharp focus but the focus is not on the eyeball itself. Many of the pictures (of animals) would be useable, but the eyeball is still not sharp. I got the same results whether I was inside 2 feet or at 10 feet--this is also a change from when the sensitivity is set at 3.
This is a very interesting result. It does seem to make it clear that the problem here is with their algorithms. The lens can acquire focus on the hairs but is not distinguishing where the eyeball is.
Yes, I had updated to version 3 of the firmware before ever using the 45mm lens. So I was getting the poor eye-af and tracking autofocus with the new firmware. I never use af-s and so I can't say anything about the performance with the update. It is interesting that Sigma is doing some firmware updating with this lens that is directed at autofocus performance. It would nice if they were working on something that would fix af-c.
p.4 #12Very frustrated by Sigma 45mm 2.8 autofocus problems
I have used the 45 with both my A7III and A7c and AF-C and my kids and I find that in good light EYE AF is spot on in almost all cases, even at close distances wide open. Only in dim light I find that it is not as accurate anymore. But the I would not use a 2.8 prime for lowlight shooting anyway. Still on FW v2, need to find a windows PC to update to v3.
p.4 #13Very frustrated by Sigma 45mm 2.8 autofocus problems
Teo Rey wrote: tedgoudie wroteJochenb wroteI still believe it's because of Sigma's mediocre AF algorithm, because all the DG DN lenses I own and tried behave more or less the same.
The very first shot in an AF-C burst is perfectly in focus most of the time. The following shots of that burst often go out of focus. Do a new burst and the first shot is great again. Etc...
So the AF motors are certainly speedy enough. IMHO the Sigma AF algorithm just isn't good at tracking a subject, but acquiring focus is not an issue.
I don't have these issues with native Sony lenses (that don't have slow AF motors).By chance, do you have the 85mm DG DN? Right now, I have the 45mm, 85mm, and 105mm macro DG DN's, shooting with an A9, typically in wide tracking mode/AF-C for everyday shooting purposes. My 45mm is not reliable in AF-C like many others have reported.
My 85mm DG DN, however, is rock solid in AF-C, both around the house and for indoor sports high frame rate stuff. The 85mm almost never misses and gets the same keeper rates as my native Sony lenses.
I've barely used the 105 macro since I got it, and I probably wouldn't use it in wide/AF-C anyway, so I've got nothing to report there. In my experience, the 45mm is not representative of my experience with other DG DN lenses.When I had the 85 DG DN, I only was able to get in focus shots about 1 or 2 out of 10 times when shooting indoors or at longer distances. I mean perfectly sharp. Usable shots were maybe more like 3-4 most of the time. I'm using the a7R III with the latest firmware.
It's such a shame too because it's such a razor sharp lens with nice rendering when it lands, but in the scenarios I need a fast f1.4 portrait lens (full length portraits at a distance or high ISO indoor shooting) it just failed pretty miserably. I was super surprised by its bad performance considering I also use the 28mm f1.4 ART which is originally a DSLR lens and AF seemed to perform better on that. Although it's possible I was using spot focus area more and eye AF less with that lens so that could be why. *Edit* yes, now that I remember I do believe I would often use spot focus when more than let's say 5 or 6 feet or so away with the 28mm, as eye AF would be inconsistent at further distances.
Maybe I just got a dud copy of the 85, I don't know. I never tested another copy. Right now I'm looking into the 65, as I prefer the rendering even more, although I really really wish it was f1.4.
*Edit* I should mention that my shots are mostly all in single shot drive mode, so none of these results were obtained while shooting and tracking a subject simultaneously. When shooting portraits, I'll hold down eye AF and take one shot at a time as I'm usually using flash or I just don't want to save too many to my SD card. I never knew about the Blue Bomber Turbo "hack" so my tracking sensitivity settings were at default values when shooting with the 85 and the 28mm....Show more →
This is really strange to hear. In the scenario you describe I find it hard to distinguish the Sigma from fast native Sony lens...
p.4 #14Very frustrated by Sigma 45mm 2.8 autofocus problems
j4nu wrote: Teo Rey wrotetedgoudie wroteJochenb wroteI still believe it's because of Sigma's mediocre AF algorithm, because all the DG DN lenses I own and tried behave more or less the same.
The very first shot in an AF-C burst is perfectly in focus most of the time. The following shots of that burst often go out of focus. Do a new burst and the first shot is great again. Etc...
So the AF motors are certainly speedy enough. IMHO the Sigma AF algorithm just isn't good at tracking a subject, but acquiring focus is not an issue.
I don't have these issues with native Sony lenses (that don't have slow AF motors).By chance, do you have the 85mm DG DN? Right now, I have the 45mm, 85mm, and 105mm macro DG DN's, shooting with an A9, typically in wide tracking mode/AF-C for everyday shooting purposes. My 45mm is not reliable in AF-C like many others have reported.
My 85mm DG DN, however, is rock solid in AF-C, both around the house and for indoor sports high frame rate stuff. The 85mm almost never misses and gets the same keeper rates as my native Sony lenses.
I've barely used the 105 macro since I got it, and I probably wouldn't use it in wide/AF-C anyway, so I've got nothing to report there. In my experience, the 45mm is not representative of my experience with other DG DN lenses.When I had the 85 DG DN, I only was able to get in focus shots about 1 or 2 out of 10 times when shooting indoors or at longer distances. I mean perfectly sharp. Usable shots were maybe more like 3-4 most of the time. I'm using the a7R III with the latest firmware.
It's such a shame too because it's such a razor sharp lens with nice rendering when it lands, but in the scenarios I need a fast f1.4 portrait lens (full length portraits at a distance or high ISO indoor shooting) it just failed pretty miserably. I was super surprised by its bad performance considering I also use the 28mm f1.4 ART which is originally a DSLR lens and AF seemed to perform better on that. Although it's possible I was using spot focus area more and eye AF less with that lens so that could be why. *Edit* yes, now that I remember I do believe I would often use spot focus when more than let's say 5 or 6 feet or so away with the 28mm, as eye AF would be inconsistent at further distances.
Maybe I just got a dud copy of the 85, I don't know. I never tested another copy. Right now I'm looking into the 65, as I prefer the rendering even more, although I really really wish it was f1.4.
*Edit* I should mention that my shots are mostly all in single shot drive mode, so none of these results were obtained while shooting and tracking a subject simultaneously. When shooting portraits, I'll hold down eye AF and take one shot at a time as I'm usually using flash or I just don't want to save too many to my SD card. I never knew about the Blue Bomber Turbo "hack" so my tracking sensitivity settings were at default values when shooting with the 85 and the 28mm.This is really strange to hear. In the scenario you describe I find it hard to distinguish the Sigma from fast native Sony lens...
It's always possible it was user error or a bad copy, I suppose. What settings do you recommend using?
p.4 #15Very frustrated by Sigma 45mm 2.8 autofocus problems
Teo Rey wrote: j4nu wroteTeo Rey wrotetedgoudie wroteJochenb wroteI still believe it's because of Sigma's mediocre AF algorithm, because all the DG DN lenses I own and tried behave more or less the same.
The very first shot in an AF-C burst is perfectly in focus most of the time. The following shots of that burst often go out of focus. Do a new burst and the first shot is great again. Etc...
So the AF motors are certainly speedy enough. IMHO the Sigma AF algorithm just isn't good at tracking a subject, but acquiring focus is not an issue.
I don't have these issues with native Sony lenses (that don't have slow AF motors).By chance, do you have the 85mm DG DN? Right now, I have the 45mm, 85mm, and 105mm macro DG DN's, shooting with an A9, typically in wide tracking mode/AF-C for everyday shooting purposes. My 45mm is not reliable in AF-C like many others have reported.
My 85mm DG DN, however, is rock solid in AF-C, both around the house and for indoor sports high frame rate stuff. The 85mm almost never misses and gets the same keeper rates as my native Sony lenses.
I've barely used the 105 macro since I got it, and I probably wouldn't use it in wide/AF-C anyway, so I've got nothing to report there. In my experience, the 45mm is not representative of my experience with other DG DN lenses.When I had the 85 DG DN, I only was able to get in focus shots about 1 or 2 out of 10 times when shooting indoors or at longer distances. I mean perfectly sharp. Usable shots were maybe more like 3-4 most of the time. I'm using the a7R III with the latest firmware.
It's such a shame too because it's such a razor sharp lens with nice rendering when it lands, but in the scenarios I need a fast f1.4 portrait lens (full length portraits at a distance or high ISO indoor shooting) it just failed pretty miserably. I was super surprised by its bad performance considering I also use the 28mm f1.4 ART which is originally a DSLR lens and AF seemed to perform better on that. Although it's possible I was using spot focus area more and eye AF less with that lens so that could be why. *Edit* yes, now that I remember I do believe I would often use spot focus when more than let's say 5 or 6 feet or so away with the 28mm, as eye AF would be inconsistent at further distances.
Maybe I just got a dud copy of the 85, I don't know. I never tested another copy. Right now I'm looking into the 65, as I prefer the rendering even more, although I really really wish it was f1.4.
*Edit* I should mention that my shots are mostly all in single shot drive mode, so none of these results were obtained while shooting and tracking a subject simultaneously. When shooting portraits, I'll hold down eye AF and take one shot at a time as I'm usually using flash or I just don't want to save too many to my SD card. I never knew about the Blue Bomber Turbo "hack" so my tracking sensitivity settings were at default values when shooting with the 85 and the 28mm.This is really strange to hear. In the scenario you describe I find it hard to distinguish the Sigma from fast native Sony lens...It's always possible it was user error or a bad copy, I suppose. What settings do you recommend using?
I don't really have very specific settings, 99% of the time I shoot AF-C with default sensitivity, mostly with eyeAF.
Regardless if closer or further away I find 85DN focus accuracy rather high, except for sudden movements or when the target moves towards me in closer distance. Maybe the glass is too heavy for the stepper motor to move it fast enough in such cases... When it comes to movement from one side of the frame to the other (e.g. left to right), it's rather good I'd say - not lightning fast but not slow either...
p.4 #17Very frustrated by Sigma 45mm 2.8 autofocus problems
scalanc2 wrote:
Coming back to the subject I see no improvement with the new firmware in AFC while on AFS il is just marginal.
For me AF-S was already accurate before the update. AF-C might be a bit better, but hard to tell for sure as in my case the issue really only exists when shooting wide open *and* at close distances. Further away I think DoF masks most focusing inaccuracies...
p.4 #18Very frustrated by Sigma 45mm 2.8 autofocus problems
j4nu wrote: Teo Rey wrotej4nu wroteTeo Rey wrotetedgoudie wroteJochenb wroteI still believe it's because of Sigma's mediocre AF algorithm, because all the DG DN lenses I own and tried behave more or less the same.
The very first shot in an AF-C burst is perfectly in focus most of the time. The following shots of that burst often go out of focus. Do a new burst and the first shot is great again. Etc...
So the AF motors are certainly speedy enough. IMHO the Sigma AF algorithm just isn't good at tracking a subject, but acquiring focus is not an issue.
I don't have these issues with native Sony lenses (that don't have slow AF motors).By chance, do you have the 85mm DG DN? Right now, I have the 45mm, 85mm, and 105mm macro DG DN's, shooting with an A9, typically in wide tracking mode/AF-C for everyday shooting purposes. My 45mm is not reliable in AF-C like many others have reported.
My 85mm DG DN, however, is rock solid in AF-C, both around the house and for indoor sports high frame rate stuff. The 85mm almost never misses and gets the same keeper rates as my native Sony lenses.
I've barely used the 105 macro since I got it, and I probably wouldn't use it in wide/AF-C anyway, so I've got nothing to report there. In my experience, the 45mm is not representative of my experience with other DG DN lenses.When I had the 85 DG DN, I only was able to get in focus shots about 1 or 2 out of 10 times when shooting indoors or at longer distances. I mean perfectly sharp. Usable shots were maybe more like 3-4 most of the time. I'm using the a7R III with the latest firmware.
It's such a shame too because it's such a razor sharp lens with nice rendering when it lands, but in the scenarios I need a fast f1.4 portrait lens (full length portraits at a distance or high ISO indoor shooting) it just failed pretty miserably. I was super surprised by its bad performance considering I also use the 28mm f1.4 ART which is originally a DSLR lens and AF seemed to perform better on that. Although it's possible I was using spot focus area more and eye AF less with that lens so that could be why. *Edit* yes, now that I remember I do believe I would often use spot focus when more than let's say 5 or 6 feet or so away with the 28mm, as eye AF would be inconsistent at further distances.
Maybe I just got a dud copy of the 85, I don't know. I never tested another copy. Right now I'm looking into the 65, as I prefer the rendering even more, although I really really wish it was f1.4.
*Edit* I should mention that my shots are mostly all in single shot drive mode, so none of these results were obtained while shooting and tracking a subject simultaneously. When shooting portraits, I'll hold down eye AF and take one shot at a time as I'm usually using flash or I just don't want to save too many to my SD card. I never knew about the Blue Bomber Turbo "hack" so my tracking sensitivity settings were at default values when shooting with the 85 and the 28mm.This is really strange to hear. In the scenario you describe I find it hard to distinguish the Sigma from fast native Sony lens...It's always possible it was user error or a bad copy, I suppose. What settings do you recommend using?I don't really have very specific settings, 99% of the time I shoot AF-C with default sensitivity, mostly with eyeAF.
Regardless if closer or further away I find 85DN focus accuracy rather high, except for sudden movements or when the target moves towards me in closer distance. Maybe the glass is too heavy for the stepper motor to move it fast enough in such cases... When it comes to movement from one side of the frame to the other (e.g. left to right), it's rather good I'd say - not lightning fast but not slow either...
Hmm, that's what I typically do too but that's where I was having issues. Default sensitivity, Eye AF, AF priority, AF-C, single shot. All on the a7R III.
p.4 #19Very frustrated by Sigma 45mm 2.8 autofocus problems
Teo Rey wrote: j4nu wroteTeo Rey wrotej4nu wroteTeo Rey wrotetedgoudie wroteBy chance, do you have the 85mm DG DN? Right now, I have the 45mm, 85mm, and 105mm macro DG DN's, shooting with an A9, typically in wide tracking mode/AF-C for everyday shooting purposes. My 45mm is not reliable in AF-C like many others have reported.
My 85mm DG DN, however, is rock solid in AF-C, both around the house and for indoor sports high frame rate stuff. The 85mm almost never misses and gets the same keeper rates as my native Sony lenses.
I've barely used the 105 macro since I got it, and I probably wouldn't use it in wide/AF-C anyway, so I've got nothing to report there. In my experience, the 45mm is not representative of my experience with other DG DN lenses.When I had the 85 DG DN, I only was able to get in focus shots about 1 or 2 out of 10 times when shooting indoors or at longer distances. I mean perfectly sharp. Usable shots were maybe more like 3-4 most of the time. I'm using the a7R III with the latest firmware.
It's such a shame too because it's such a razor sharp lens with nice rendering when it lands, but in the scenarios I need a fast f1.4 portrait lens (full length portraits at a distance or high ISO indoor shooting) it just failed pretty miserably. I was super surprised by its bad performance considering I also use the 28mm f1.4 ART which is originally a DSLR lens and AF seemed to perform better on that. Although it's possible I was using spot focus area more and eye AF less with that lens so that could be why. *Edit* yes, now that I remember I do believe I would often use spot focus when more than let's say 5 or 6 feet or so away with the 28mm, as eye AF would be inconsistent at further distances.
Maybe I just got a dud copy of the 85, I don't know. I never tested another copy. Right now I'm looking into the 65, as I prefer the rendering even more, although I really really wish it was f1.4.
*Edit* I should mention that my shots are mostly all in single shot drive mode, so none of these results were obtained while shooting and tracking a subject simultaneously. When shooting portraits, I'll hold down eye AF and take one shot at a time as I'm usually using flash or I just don't want to save too many to my SD card. I never knew about the Blue Bomber Turbo "hack" so my tracking sensitivity settings were at default values when shooting with the 85 and the 28mm.This is really strange to hear. In the scenario you describe I find it hard to distinguish the Sigma from fast native Sony lens...It's always possible it was user error or a bad copy, I suppose. What settings do you recommend using?I don't really have very specific settings, 99% of the time I shoot AF-C with default sensitivity, mostly with eyeAF.
Regardless if closer or further away I find 85DN focus accuracy rather high, except for sudden movements or when the target moves towards me in closer distance. Maybe the glass is too heavy for the stepper motor to move it fast enough in such cases... When it comes to movement from one side of the frame to the other (e.g. left to right), it's rather good I'd say - not lightning fast but not slow either...Hmm, that's what I typically do too but that's where I was having issues. Default sensitivity, Eye AF, AF priority, AF-C, single shot. All on the a7R III.
I don't know how to explain it then . Maybe you're more critical of sharpness than me... I can post some rather long distance snaps from yesterdays walk...
p.4 #20Very frustrated by Sigma 45mm 2.8 autofocus problems
I updated to the latest firmware last week and it didn't seem to improve AF at all. I stopped using AFC with this lens a while ago and mainly shoot it now on AFS.
I shot exclusively with this lens yesterday and I have a strong love hate relationship with this lens. It works consistently with grown ups but with my toddler I maybe only got 60% of shots in focus. In some shots I had to add clarity to make it acceptably sharp which screwed up the bokeh. My wife's hit rate is even lower and using this lens yesterday was mostly frustrating. Then there are days when I go out to shoot with the lens and images knocks the socks off. As of right now, i feel like i should just let it go. Nobody should suffer with this kind of AF especially on Sony's AF tech.
I need something with 35/2.8 ZA like AF performance and 45/2.8 like rendering. Sigma 35I might do the trick, but it is bigger and heavier 😞.