PDX10 "Film Look"

Shawn Mielke · February 2nd, 2004, 08:23 PM

Sorry, Evan, didn't mean to mislead. The PDX10's 4:3 image is as lustrous as the 16:9. Boyd helps make a key point, though. I don't know the GL2's image, so I can't help you understand what I know in terms of what you know. I can repeat what others have said, and that is that, in good light, the PDX10s image is comparable, and even sometimes preferrable, to the pd150's image, if that helps in the least. Between that and Boyd's comments about the vx2000, it (the PDX10) seems to be more or less in league with the much praised prosumer Sony 1/3" chip models.

ps Two years ago, I used the XL1S for one week with the standard lens. It was my first experience with DV, but I liked the images it produced very much....

Scott Plowman · February 4th, 2004, 06:00 PM

Lets remind Evan that the PDx10 is a 1/4 chipper not a 1/3 chipper.. Beautiful pic?? Yes very much so..
The crux is this.. Are we doing weddings or other low light work??? If this is so we need something that has a much higher low light ability.. its all in the lux.. Thanks

Boyd Ostroff · February 4th, 2004, 07:24 PM

Scott, earlier you said:

Quote:

It has issues with low light ability.. abysmal as its been described.. I have seen posts on the net of comparison frames.. It really is bad at low light.. I mean unusable.

Just curious, do you base this on personal experience with the PDX-10, or on comments from others? There is no question that the low light abilities of the VX-2x00 and PD-1x0 are better. However I don't find the PDX-10 "unusable" in low light by any means. You can shoot with the gain up pretty high and still get a decent image due to the DXP chips. I just shot video of a REALLY dark opera with the PDX-10. I was wide open with the gain anywhere between 3dB and 12dB most of the time. But it really doesn't look so bad to my eyes. I've shot video by campfire light, and at night on the beach with only a flashlight as fill. There's noise, but again it looks pretty good to me. But of course I was going for more of an "arsty look" and not a documentary.

Now if you are really shooting most of your stuff under candle light or outside at night I'd agree you might not want a PDX-10. All along I've maintained that the primary question you need to answer for yourself is "how important is 16:9". Unless it's very important to you then there probably isn't a compelling reason to get the PDX-10. But if it's very important to you then I suspect you'll learn to live with its shortcomings. And for that matter, where is it written that one camera is the correct choice for all your needs? If I'm really worried about low light levels that's what my VX-2000 is for (although I use it primarily as a deck these days :-)

Evan Kubota · February 5th, 2004, 11:08 PM

"Lets remind Evan that the PDx10 is a 1/4 chipper"

If you want to get technical, it's 1/4.7"... I have read that it's a full stop less sensitive than the GL2 in low light conditions. I don't intend to shoot in terrible lighting, but nonetheless, I prefer to use as little gain as possible, so I decided to go with the GL2. It wasn't just about the low-light though - I also wanted the control over the ND filters, larger CCDs, and 20x zoom. However, frame mode was the most important. I'd rather not have to de-interlace every single second of footage in post-production. It was a tough choice because I liked the PDX-10's audio capabilities, more compact form factor, and lighter weight. Native 16:9 would also have been fun. However, given that most of my work is destined for 4:3 ratio TVs, etc., I made the decision to go for the GL2, shoot in 4:3 with 16:9 framing guides, and letterbox in post.

Thanks for the help, everyone.

Shawn Mielke · February 6th, 2004, 11:42 AM

The GL2 is 6.5 oz heavier. I guess that could become substantial by the end of the day.

The PDX10 chips do a remarkable job of suppressing noise. Don't know how how well Canon chips handle this.

Ignacio Rodriguez · February 6th, 2004, 01:56 PM

Not having ever been a 'film' guy, I am not after such look per se, however I find that there are things I can do to DV which give it a texture I like better than raw video and the result does seem to look like film transferred to video to some extent. This is what I use:

PDX10 in 16x9 mode, sometimes with 1/30 shutter. *
Sony 0.7 WA adapter. It softens the image a bit.
Set edge enhancement ('sharpening') to a minimum (custom preset).
Set chroma ('color level') to somewhat less than default (custom preset).
Deinterlace in FCP.
Desaturate highlights and lowlights in FCP.
Use Stib's Film Curves 1.0 (slightly) and Stib's Simple Levels.
Sometimes, use Too Much Too Soon's Noise Reduction plug in FCP. **
Use FCP4's Motion Blur slightly, reduces noise and creates film-like motion.

* even though most Sony DV cameras show a visible drop in vertical resolution when going below 1/30, It seems that this camera's image processing and high-res CCD array give out a much better picture than other models, but only when using 16x9, not 4x3.

** After using Stib's curves to get more contrast from the midtones, noise becomes more visible, so I usually apply some noise reduction algorithm and/or blurring to clean it up a bit. This is probably much worse if you are using a noisier camera.

Boyd Ostroff · February 6th, 2004, 02:09 PM

<<<-- Originally posted by Ignacio Rodriguez : * even though most Sony DV cameras show a visible drop in vertical resolution when going below 1/30, It seems that this camera's image processing and high-res CCD array give out a much better picture than other models -->>>

Maybe this is just subjective? I also like that 1/30 "look", so just recently when shooting a very dark opera I tried using 1/30 sec shutter speed to gain an extra f-stop. Looking at the footage on the monitor at home, it was significantly worse than 1/60. When you think about it, seems the only way that 1/30 sec shutter speed could be implemented would be to write the same data to each of the interlaced fields. That would effectively halve the vertical resolution, and it looks this way to my eyes.

I ended up shooting my opera at 1/60 and boosting the gain a click. This really gave better results. I think you're better off using a form of adaptive deinterlacing to create 30p if you want that sort of look. Does FCP 4 do this? FCP 3 doesn't, although you can do the old trick of stacking two copies with alternate fields and making the top one 50% transparent. In FCP 3 the deinterlacer also results in throwing away half of your data, with a noticeable loss of resolution. Joe's filters has a nice deinterlacer with motion differencing, and so does DVFilm Maker. But note that there isn't any point to using these if you're shooting at 1/30, they're intended to convert 60i to 30p.

Can you tell us some more about "Stib's curves" and "Too Much Too Soon"? Haven't heard of these and they sound interesting!

Ignacio Rodriguez · February 6th, 2004, 03:52 PM

> Maybe this is just subjective? I also like that 1/30 "look",
> so just recently when shooting a very dark opera I tried
> using 1/30 sec shutter speed to gain an extra f-stop.
> Looking at the footage on the monitor at home, it was
> significantly worse than 1/60.

Perhaps you are right. The comparison I ran where I saw no difference was a quick test in a store, I don't have a 16x9 TV with S-video so the difference might be to subtle to be visible on my SDTV. I remember I could very much see the difference with my PC3 but that was a single-chip' toy...

I had always thought about doing what you mention, that is: underexposing and then using the NLE to bring the image up to the desired gain, bt have never really tried that.

FCP has some 'flicker filter' thing available when deinterlacing which seems to use both fields. Also there are some free deinterlacing filters whcih might be better than Apple's default. Stib's deinterlacer claims to be faster.

BTW, Stib's and other free plug-ins are available following this link:
http://www.digitalzoo.com.au/lunchti...ed_02_free.htm

Boyd Ostroff · February 6th, 2004, 04:42 PM

I think the monitor does explain it. I'm using a 16:9 LCD fed by component video from a DVD recorder which is connected to my Mac via firewire. The quality improvement over s-video is significant. This setup highlights the PDX-10's stengths (and weaknesses) very nicely.

Regarding the gain, I was actually refering to clicking the gain up another stop on the camera, however I sometimes manipulate the image further while editing.

Ignacio Rodriguez · February 6th, 2004, 04:49 PM

> Regarding the gain, I was actually refering to clicking the
> gain up another stop on the camera, however I
> sometimes manipulate the image further while editing.

Ohh I see... but when I use 1/30 is when light is getting so low that gain doesn't expose high enough at 1/60. If there is enough light I'll prefer 1/60... so I was usually doing the right thing but not for the same reason... for me it was because I can get smoother slow motion, which I tend to use often in documentary work.

Hmmmm... I wonder wether the PAL PDX10 besides having better res might also be slightly more sensitive to light thanks to the lower frame rate...

Boyd Ostroff · February 6th, 2004, 05:21 PM

Do you mean you get smoother slow motion at 1/30? I would have to disagree with that. 1/60 gives you more temporal data for FCP's frame blending. I've done quite a bit of this and definitely prefer 1/60. But perhaps you're going for a different look?

Ignacio Rodriguez · February 6th, 2004, 05:27 PM

> Do you mean you get smoother slow motion at 1/30?
> I would have to disagree with that. 1/60 gives you more
> temporal data for FCP's frame blending.

No, no perhaps I misexplained it. I meant that I was usually ONLY using 1/30 in the camera when there is not enough light, even for 1/60 with gain. And even though I could not see a difference in spatial quality when going down to 1/30, I preferred 1/60 becasue I would later be using slow-mo. So I was usually doing the right thing but not for the same reason.

My preferred method for getting a 'frame mode' look is by deinterlacing in FCP, not using 1/30 from the camera. I guess I did not explain that clearly.

Now regarding low light, what I have never tried to do is just let the image stay underexposed and fix it in post, which --as you stated later on-- is not exactly what you meant either. However, because of the line doubling phenomenon, the end result might be better by staying out of 1/30 as much as possible... when even the highest gain setting in the camera will not give you a correct exposure.

An interesting test would be to see whether an equivalent amount of gain is better applied in the camera or in the NLE... I wonder which result would yield less noise, better contrast, etc.

Ignacio Rodriguez · February 9th, 2004, 09:15 PM

> When you think about it, seems the only way that 1/30
> sec shutter speed could be implemented would be to write
> the same data to each of the interlaced fields.

Hmmm. I am not really sure about that being the only way... would then a 1/15 sec. shutter speed have 1/4 vertical resolution and so forth?

I am sure you know what you are talking about when you say you can see a difference on the monitor, I just think Sony could do better than that and give us real progressive scan for a better 1/30 sec shutter speed and slower.

The future is 16:9 and 30 fps. Ok, we don't get the high res yet... but we could sure use a firmware upgrade to proscan...

Ok. End of rant. Better get back to work so I can pay for my next cam, which WILL have real proscan ;-)

Boyd Ostroff · February 9th, 2004, 10:32 PM

Hmm, this stuff is confusing to me too. But I don't think you'd get 1/4 resolution if you shot at 1/15 sec. I think there would still be the same data written to each interlaced field, but it would be in two successive frames (eg: 4 fields) and with motion blur spanning two frames.

Just my educated guess based on what I know. Perhaps someone with a better technical background can explain. I like the 30p feeling too, but it just isn't a feature of this camera. And for what it's worth, if you ever wanted to go to 24p then I don't think it can be properly done from 30p footage, however it can be approximated from 60i footage.

Michael Struthers · February 12th, 2004, 01:41 AM

Buy a pdx10 PAL, get 20 pct more resolution, shoot as clean as you can and then play with the picture in post. Should look very nice.

February 2nd, 2004, 08:23 PM	#16
Shawn Mielke Major Player Join Date: Jul 2003 Location: Santa Rosa, California Posts: 745	Sorry, Evan, didn't mean to mislead. The PDX10's 4:3 image is as lustrous as the 16:9. Boyd helps make a key point, though. I don't know the GL2's image, so I can't help you understand what I know in terms of what you know. I can repeat what others have said, and that is that, in good light, the PDX10s image is comparable, and even sometimes preferrable, to the pd150's image, if that helps in the least. Between that and Boyd's comments about the vx2000, it (the PDX10) seems to be more or less in league with the much praised prosumer Sony 1/3" chip models. ps Two years ago, I used the XL1S for one week with the standard lens. It was my first experience with DV, but I liked the images it produced very much.... __________________ Breakthrough In Grey Room

February 6th, 2004, 11:42 AM	#20
Shawn Mielke Major Player Join Date: Jul 2003 Location: Santa Rosa, California Posts: 745	The GL2 is 6.5 oz heavier. I guess that could become substantial by the end of the day. The PDX10 chips do a remarkable job of suppressing noise. Don't know how how well Canon chips handle this. __________________ Breakthrough In Grey Room

February 6th, 2004, 01:56 PM	#21
Ignacio Rodriguez Contributor Join Date: Sep 2003 Location: Santiago, Chile Posts: 932	Not having ever been a 'film' guy, I am not after such look per se, however I find that there are things I can do to DV which give it a texture I like better than raw video and the result does seem to look like film transferred to video to some extent. This is what I use: PDX10 in 16x9 mode, sometimes with 1/30 shutter. * Sony 0.7 WA adapter. It softens the image a bit. Set edge enhancement ('sharpening') to a minimum (custom preset). Set chroma ('color level') to somewhat less than default (custom preset). Deinterlace in FCP. Desaturate highlights and lowlights in FCP. Use Stib's Film Curves 1.0 (slightly) and Stib's Simple Levels. Sometimes, use Too Much Too Soon's Noise Reduction plug in FCP. ** Use FCP4's Motion Blur slightly, reduces noise and creates film-like motion. * even though most Sony DV cameras show a visible drop in vertical resolution when going below 1/30, It seems that this camera's image processing and high-res CCD array give out a much better picture than other models, but only when using 16x9, not 4x3. ** After using Stib's curves to get more contrast from the midtones, noise becomes more visible, so I usually apply some noise reduction algorithm and/or blurring to clean it up a bit. This is probably much worse if you are using a noisier camera. __________________ Ignacio Rodríguez in the third world. @micronauta on Twitter. Main hardware: brain, eyes, hands.

February 6th, 2004, 03:52 PM	#23
Ignacio Rodriguez Contributor Join Date: Sep 2003 Location: Santiago, Chile Posts: 932	> Maybe this is just subjective? I also like that 1/30 "look", > so just recently when shooting a very dark opera I tried > using 1/30 sec shutter speed to gain an extra f-stop. > Looking at the footage on the monitor at home, it was > significantly worse than 1/60. Perhaps you are right. The comparison I ran where I saw no difference was a quick test in a store, I don't have a 16x9 TV with S-video so the difference might be to subtle to be visible on my SDTV. I remember I could very much see the difference with my PC3 but that was a single-chip' toy... I had always thought about doing what you mention, that is: underexposing and then using the NLE to bring the image up to the desired gain, bt have never really tried that. FCP has some 'flicker filter' thing available when deinterlacing which seems to use both fields. Also there are some free deinterlacing filters whcih might be better than Apple's default. Stib's deinterlacer claims to be faster. BTW, Stib's and other free plug-ins are available following this link: http://www.digitalzoo.com.au/lunchti...ed_02_free.htm __________________ Ignacio Rodríguez in the third world. @micronauta on Twitter. Main hardware: brain, eyes, hands.

February 6th, 2004, 04:49 PM	#25
Ignacio Rodriguez Contributor Join Date: Sep 2003 Location: Santiago, Chile Posts: 932	> Regarding the gain, I was actually refering to clicking the > gain up another stop on the camera, however I > sometimes manipulate the image further while editing. Ohh I see... but when I use 1/30 is when light is getting so low that gain doesn't expose high enough at 1/60. If there is enough light I'll prefer 1/60... so I was usually doing the right thing but not for the same reason... for me it was because I can get smoother slow motion, which I tend to use often in documentary work. Hmmmm... I wonder wether the PAL PDX10 besides having better res might also be slightly more sensitive to light thanks to the lower frame rate... __________________ Ignacio Rodríguez in the third world. @micronauta on Twitter. Main hardware: brain, eyes, hands.

February 4th, 2004, 06:00 PM	#17
Scott Plowman Regular Crew Join Date: Sep 2003 Location: Kansas City, Kansas, USA Posts: 85	Lets remind Evan that the PDx10 is a 1/4 chipper not a 1/3 chipper.. Beautiful pic?? Yes very much so.. The crux is this.. Are we doing weddings or other low light work??? If this is so we need something that has a much higher low light ability.. its all in the lux.. Thanks

February 5th, 2004, 11:08 PM	#19
Evan Kubota Regular Crew Join Date: Feb 2004 Location: Gainesville, FL Posts: 47	"Lets remind Evan that the PDx10 is a 1/4 chipper" If you want to get technical, it's 1/4.7"... I have read that it's a full stop less sensitive than the GL2 in low light conditions. I don't intend to shoot in terrible lighting, but nonetheless, I prefer to use as little gain as possible, so I decided to go with the GL2. It wasn't just about the low-light though - I also wanted the control over the ND filters, larger CCDs, and 20x zoom. However, frame mode was the most important. I'd rather not have to de-interlace every single second of footage in post-production. It was a tough choice because I liked the PDX-10's audio capabilities, more compact form factor, and lighter weight. Native 16:9 would also have been fun. However, given that most of my work is destined for 4:3 ratio TVs, etc., I made the decision to go for the GL2, shoot in 4:3 with 16:9 framing guides, and letterbox in post. Thanks for the help, everyone.

February 6th, 2004, 02:09 PM	#22
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	<<<-- Originally posted by Ignacio Rodriguez : * even though most Sony DV cameras show a visible drop in vertical resolution when going below 1/30, It seems that this camera's image processing and high-res CCD array give out a much better picture than other models -->>> Maybe this is just subjective? I also like that 1/30 "look", so just recently when shooting a very dark opera I tried using 1/30 sec shutter speed to gain an extra f-stop. Looking at the footage on the monitor at home, it was significantly worse than 1/60. When you think about it, seems the only way that 1/30 sec shutter speed could be implemented would be to write the same data to each of the interlaced fields. That would effectively halve the vertical resolution, and it looks this way to my eyes. I ended up shooting my opera at 1/60 and boosting the gain a click. This really gave better results. I think you're better off using a form of adaptive deinterlacing to create 30p if you want that sort of look. Does FCP 4 do this? FCP 3 doesn't, although you can do the old trick of stacking two copies with alternate fields and making the top one 50% transparent. In FCP 3 the deinterlacer also results in throwing away half of your data, with a noticeable loss of resolution. Joe's filters has a nice deinterlacer with motion differencing, and so does DVFilm Maker. But note that there isn't any point to using these if you're shooting at 1/30, they're intended to convert 60i to 30p. Can you tell us some more about "Stib's curves" and "Too Much Too Soon"? Haven't heard of these and they sound interesting!

February 6th, 2004, 04:42 PM	#24
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	I think the monitor does explain it. I'm using a 16:9 LCD fed by component video from a DVD recorder which is connected to my Mac via firewire. The quality improvement over s-video is significant. This setup highlights the PDX-10's stengths (and weaknesses) very nicely. Regarding the gain, I was actually refering to clicking the gain up another stop on the camera, however I sometimes manipulate the image further while editing.

February 6th, 2004, 05:21 PM	#26
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	Do you mean you get smoother slow motion at 1/30? I would have to disagree with that. 1/60 gives you more temporal data for FCP's frame blending. I've done quite a bit of this and definitely prefer 1/60. But perhaps you're going for a different look?

February 6th, 2004, 05:27 PM	#27
Ignacio Rodriguez Contributor Join Date: Sep 2003 Location: Santiago, Chile Posts: 932	> Do you mean you get smoother slow motion at 1/30? > I would have to disagree with that. 1/60 gives you more > temporal data for FCP's frame blending. No, no perhaps I misexplained it. I meant that I was usually ONLY using 1/30 in the camera when there is not enough light, even for 1/60 with gain. And even though I could not see a difference in spatial quality when going down to 1/30, I preferred 1/60 becasue I would later be using slow-mo. So I was usually doing the right thing but not for the same reason. My preferred method for getting a 'frame mode' look is by deinterlacing in FCP, not using 1/30 from the camera. I guess I did not explain that clearly. Now regarding low light, what I have never tried to do is just let the image stay underexposed and fix it in post, which --as you stated later on-- is not exactly what you meant either. However, because of the line doubling phenomenon, the end result might be better by staying out of 1/30 as much as possible... when even the highest gain setting in the camera will not give you a correct exposure. An interesting test would be to see whether an equivalent amount of gain is better applied in the camera or in the NLE... I wonder which result would yield less noise, better contrast, etc. __________________ Ignacio Rodríguez in the third world. @micronauta on Twitter. Main hardware: brain, eyes, hands.

February 9th, 2004, 09:15 PM	#28
Ignacio Rodriguez Contributor Join Date: Sep 2003 Location: Santiago, Chile Posts: 932	> When you think about it, seems the only way that 1/30 > sec shutter speed could be implemented would be to write > the same data to each of the interlaced fields. Hmmm. I am not really sure about that being the only way... would then a 1/15 sec. shutter speed have 1/4 vertical resolution and so forth? I am sure you know what you are talking about when you say you can see a difference on the monitor, I just think Sony could do better than that and give us real progressive scan for a better 1/30 sec shutter speed and slower. The future is 16:9 and 30 fps. Ok, we don't get the high res yet... but we could sure use a firmware upgrade to proscan... Ok. End of rant. Better get back to work so I can pay for my next cam, which WILL have real proscan ;-) __________________ Ignacio Rodríguez in the third world. @micronauta on Twitter. Main hardware: brain, eyes, hands.

February 9th, 2004, 10:32 PM	#29
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	Hmm, this stuff is confusing to me too. But I don't think you'd get 1/4 resolution if you shot at 1/15 sec. I think there would still be the same data written to each interlaced field, but it would be in two successive frames (eg: 4 fields) and with motion blur spanning two frames. Just my educated guess based on what I know. Perhaps someone with a better technical background can explain. I like the 30p feeling too, but it just isn't a feature of this camera. And for what it's worth, if you ever wanted to go to 24p then I don't think it can be properly done from 30p footage, however it can be approximated from 60i footage.

February 12th, 2004, 01:41 AM	#30
Michael Struthers Major Player Join Date: Jul 2003 Location: SF, Ca Posts: 421	Buy a pdx10 PAL, get 20 pct more resolution, shoot as clean as you can and then play with the picture in post. Should look very nice.