CompuServe Thread

#3DSr3 corrupted *.GIF

16 messages in this thread
#130439From: T-Spect LeOct 21, 1994 2:29 PM
>>You have to do it the latter way… render the targas out and then compile them into a flic. There's no way to retain the temp files. – G<< While on this topic, I'm having constant problem when VP rendering FLC or FLI to *.GIF. On all of my four diverse PCs, they give me same problem: Most seq. #'d *.GIFs are corrupted. (I like to render to *.GIF because it saved huge amount of diskspace over the Targas) . Another thing I have to do is to re-rendered the finished seq # *.GIF [in Vpost] without saving to disk so I could verify the integrity of the file. Is this fixed on r4 or not yet? (I'm anxious to get r4 very soon) – SPECT
#130448From: Jonas Ruikis [ADESK]Oct 21, 1994 2:44 PM
Hi Spect, << Another thing I have to do is to re-rendered the finished seq # *.GIF [in Vpost] without saving to disk so I could verify the integrity of the file. Is this fixed on r4 or not yet? (I'm anxious to get r4 very soon) >> This is not a problem that is known about by me.. Send me the project file and the resultant gifs to review. What is corrupt about the gifs? jonas[adesk]
#130563From: T-Spect LeOct 22, 1994 6:05 AM
>>Send me the project file and the resultant gifs to review. What is corrupt about the gifs?<< Thanks for the concern. I'm sending you via E-mail the zip'd file called "JR-BUG1.ZIP". It contained my "AF-SC-NF.prj", "SSAR-LVE.prj" files. These were the VPost test file of all of the seq #'d GIFs. It have numerous entry & map-paths; "ASOFT-B1.prj" is the actual project file recreated for this. It have only one entry on the queue. Then you'll see "CHOP0177.gif, CHOP0178.gif, CHOP0179, CHOP0504.gif, CHOP0505.gif, CHOP0506.gif. The corrupted ones are: CHOP0178.gif, CHOP0505.gif. The rest of the GIFs are perfect. I want to give this so that you can compare byte size, ect… This was re-rendered & recreated today for the purpose of showing you the problem. There are a total of 600 frames, out of the whole, only two are corrupted. I don't think you can test any of the uploaded *.PRJ files since it's not feasable to upload more than 1 gig of the required map files. Rather, it's for you to see how it was done. || History of the problem: The original source were off the CD-ROMs, HDs. They're perfect. I'd view them via the VIEW FLIC in 3DSr3. I then convert these FLI/FLC to the *.GIF. This when the problem start. Some of these are fine, but for strange reasons, some are corrupted. This I mean that anywhere any of those file will have a bad *.GIF file. All of these rendering were done on 4 PCs [not simultanously because of the hardware lock & I don't yet have Novel network set-up]. The PCs are very diverse. They have IDE, SCSI-II; PCI, VLB, & ISA buses. RAM ranges from 32 Mb to 128 Mb. All PCs were thoroughly diagnosed & all are ok. [It's not accurate to pre-conclude that the problem is of the hardware because it's evidented that this is happening on all of the four PCs] However, all of these computer uses SMARTDRV.exe driver as a software based caching to enhances disk performance. AUTOEXEC.bat: Typically regular, no special or abnormal drivers used or loaded. CONFIG.sys: " " " " || To reproduce this problem: Find any FLI/FLC sources & render them as *.GIF [source files should be at least 80 frames]. Try several of them (5 or more]. The easiest way to verify the integrity of the files: Add the new map-path location where you'd saved them. Go to VPost and put the 1st four character of the seq#'d names plus an " * ". Then press the ALT key on VPost to set the correct range of frame. Do this for all of your rendered seq.#'d GIFs. You don't have to move the bars in VPost where its indicating when the bitmap file should start. This might not be accurate if two corrupted files coincidently happened at the same frame. My orig. PRJ file also have the BEEP.IXP ipas [shareware]. This is a handy sound feature that I used to make an audible beep as this inevitably boring test progresses] Render them without saving to disk. I bet you'll see the message: "Bitmap xxxx####*.GIF is corrupted". If none of them aren't corrupted, then I just don't know what to say. |||||| Side notes: This problem also occured on regular mesh scenes by rendering without VPost. Try that if your VPost passed all of the file integrity test mentioned above.||||||| || Reasonings for doing this: My main reason for doing to GIFs. 1) Significantly smaller disk space consumption. 2) The IFL (image file list) format only takes the 1st frame of the FLC, so I must use the seq. # frames to have more than one sequence of animated maps. I just don't want several gigs of composited FLCs where w/ IFL, I could select & re-use the component files for my scene, thus requiring much lesser disk space to hold them. || If this problem is not solved, then I have to alway verify the integrity of the GIFs by using VPost. Countless hours spending on putting on the queue & waiting for the corrupted file to appear. Then I have to take note on those pesky corrupted files. This could take days. Then re-render them again to make sure they're ok. When they aren't, I have to go to XTree Gold 3.0 to copy the previous good one & renamed to the next sequentially # frames. What a pain. I have yet to test the TIFFs, TGAs, BMPs, JPGs ! ! ! ! Regards, SPECT
#130493From: Yost GroupOct 21, 1994 6:14 PM
As Jonas says, there are no open problems in our gif reading/writing code. Those routines are completely stable, so if you're getting corrupted files, you need to start looking carefully at your hardware. – G
#130532From: T-Spect LeOct 21, 1994 9:47 PM
>>you need to start looking carefully at your hardware.<< Gary: Yes, I t h o r o u g h l y diagnosed my 4 PCs using QA Plus, MSD, and other softwares, plus scanning all of my IDE & SCSI-II hard drives, and there are no problems. However, all of my PCs are using the SMARTDRV.exe . I couldn't find any documentation(s) in the 3DS's Installation Guide about the Smartdrive driver. I followed all of the instructions (not to use Memmaker, use HIMEM.sys, ect…) I also didn't modify any of the CFIG386.exe settings. Most of these FLI/FLC to *.GIF conversions in VPost took only 1 sec/frame on the Pentium PCs. I noticed a 6-10 second delay between completed rendering & writting to disk cycle. D o y o u think the SMARTDRV.exe might c a u s e t h e c o r r u p t i o n to the *.GIF ? [I have yet to test without using the smartdrv.exe driver] – SPECT
#130545From: Jonas Ruikis [ADESK]Oct 21, 1994 11:04 PM
T-Spect Le, << D o y o u think the SMARTDRV.exe might c a u s e t h e c o r r u p t i o n to the *.GIF ? >> There have been successful users of smartdrv over the last year and there have been some that were not. In a production environment, some would rather invest in more hardware rather than use additional compression techniques. 3DS users tend to tax their environments and an intermittant problem is hard to tolerate when it affects production. Trying to diagnose a failing machine is also difficult when you add the layer of possible errors during compression. I have heard of individuals who compress the drive that they keep the system's static files and executables. However, these individuals keep the swap file and images on an uncompressed drive. This seems to balance the two naturally. jonas[adesk]
#130564From: T-Spect LeOct 22, 1994 6:06 AM
Hi Jonas, >>There have been successful users of smartdrv over the last year and there have been some that were not. In a production environment, some would rather invest in more hardware rather than use additional compression techniques. 3DS users tend to tax their environments and an intermittant problem is hard to tolerate when it affects production. Trying to diagnose a failing machine is also difficult when you add the layer of possible errors during compression. << None of my PC hard drives are compressed, which is DOUBLESPACING. From your message [correct me if I'm wrong], I think you're referring SMARTDRV.exe as disk compression. It's a RAM to disk caching method. I'm not fully sure if SMARTDRV.exe uses compression during the cache cycle. Anyhow, I did try to DOUBLESPACE my hard drive last year. The results were so "traumatic" that I vowed never to use it again. – SPECT
#130876From: SyndesisOct 24, 1994 10:16 AM
When SmartDrive dumps its buffer, it can interfere with other devices (like serial ports) because of the subsequent delays in processing interrupts. Remove it and see if it makes a difference.
#130590From: Yost GroupOct 22, 1994 11:21 AM
The bottom line is that, out of almost 40,000 licensed 3DS users, you are the only person who has ever reported a problem with gif writing. You've sent Jonas a project file for him to look at, and he'll start on it to see if reproducing it is possible. Re smartdrive… a cache on certain machines could certainly cause the problem. Try uninstalling it and see what happens. – G
#131819From: Volker UffelmannOct 28, 1994 9:25 AM
>> The bottom line is that, out of almost 40,000 licensed 3DS users, you are the only person who has ever reported a problem with gif writing. << Well, not really. You probably don't remember that I also have complained about corrupted GIFs… and in the meantime it looks as if some of my TIFs are corrupted too… Some lines in an image are damaged (showing a colorful variety of beautiful pixels <g>). It is definitely _NOT_ the hardware (thorough checks of the data integrity of the used harddisks/controllers/networks showed no errors – no smartdrv on the network) – so what's left? Interesting anyway to learn that someone else encounters the same problems as I do – one could think that there's more to it than pure coincidence…? Volker PS: The errors show up not only during VideoPost but also during "normal" Keyframer renderings (one out of 200 images or so) – haven't seen it happen with a 3D modeler rendering (but usually I don't render some hundred pictures from the 3D modeler…)
#131826From: Yost GroupOct 28, 1994 9:57 AM
If you're getting a "colorful variety of beautiful pixels" in your gif and/or tif files, I personally guarantee you that you've got some sort of hardware problem, or memory management incompatibility. This problems is similar to the "Invalid .3DS (or .PRJ) file" corruption that some folks get. I've hard of problems with .TGA file writing as well, when the HD controller had a slight incompatibility to Phar Lap. You need to look deeper into your hardware configuraton to find the problem. – G
#131871From: MARTIN G FOSTEROct 28, 1994 1:19 PM
Volker, >> Some lines in an image are damaged (showing a colorful variety of beautiful pixels <g>). It is definitely _NOT_ the hardware << I've had similar problems in the past, with the occasional corrupted .tga file – and it turned out to be nothing to do with the software. I did some checking of my network, replaced my network cards and upgraded from Lantastic 4.0 to 5.0 and the problem went away. While I sympathize with your problem, I would bet money that it IS your hardware or network. It certainly was in my case.
#131884From: David J. MarksOct 28, 1994 2:11 PM
>> Some lines in an image are damaged (showing a colorful variety of >> beautiful pixels <g>) I had a similar problem with JPG's that were related to a bad hard drive. I performed a surface scan on the disk and found several bad clusters. Once the bad sectors were locked out, the problem went away. I ran into another problem displaying files on a Targa board because of a memory conflict. Excluding D000-DFFF cleared up that problem. Yet another problem was traced back to a faulty network card that corrupted image files on their way to the server.
#130922From: Forcade & AssociatesOct 24, 1994 1:00 PM
SPECT, I don't know about your corrupt GIF file problem. If I were you I would run some tests without smartdrive to see if it was part of the problem. As an aside, you might get better quality with similar compression ratios using compressed TIFFs instead of GIFs. GIFs are reduced color pallette dithered files. When you render with them they must be redithered and repalletted. This can cause visual artifacts. TIFFs use a similar compression scheme as GIFs but they are 24-bit files. They may not compress as much as GIFs but they will compress much more than TGA files do. I always try to keep as much image quality as possible as far into the process as I can. Terry Gilbert Forcade & Associates
#131421From: T-Spect LeOct 26, 1994 12:48 PM
Terry: Thanks for the reply to the GIF problem. >>As an aside, you might get better quality with similar compression ratios using compressed TIFFs instead of GIFs.<< Have you try to render in 3DSr3 to *.TIF or *.TGA w i t h o u t using the "compression" option in 3DS? I found out that the majority of the time by using PKZIP compression software, you get almost 46% ratio than with the LZW compression (20-30% ratio) used in TIFFs (I don't know what Targa uses) THIS VARIES from image to image. SPECT
#131881From: Forcade & AssociatesOct 28, 1994 1:57 PM
>> Terry: Thanks for the reply to the GIF problem. >>As an aside, you might get better quality with similar compression ratios using compressed TIFFs instead of GIFs.<< Have you try to render in 3DSr3 to *.TIF or *.TGA w i t h o u t using the "compression" option in 3DS? I found out that the majority of the time by using PKZIP compression software, you get almost 46% ratio than with the LZW compression (20-30% ratio) used in TIFFs (I don't know what Targa uses) THIS VARIES from image to image. SPECT << Yes the compression does vary from image to image. TGA compression is a simple RLE scheme that works best for flat areas. Any patterns will defeat the compression. I'm not sure which compression 3DS uses for its TIFs (packed pixel maybe). The LZW compression that Photoshop uses is basically as good as the LZ variant the Pkzip uses. The advantage to using compressed TIFs (of whatever compression scheme) is that they maintain 24-bit color and the compression is reasonably good if not always the best. JPEGs are usually the best compression but the artifacts can be a problem. For achival/transport purposes PKzip is a good choice but if you are going to be working with the files then the less compact compressed TIFs are probably better. You will have to decide on which compromise of quality, size and convenience is best for you. Terry Gilbert Forcade & Associates